Cable communication systems provide one or more of commercial TV services, Internet data services, and voice services (e.g., “Voice-over-Internet Protocol,” or VoIP) to one or more subscriber premises (or “end users”) in a given geographic area. Generally speaking, a cable communication system refers to the operational (e.g., geographical) footprint of an entertainment and/or information services franchise that provides entertainment and/or information services to a subscriber base spanning one or more towns, a metropolitan area, or a portion thereof. Particular entertainment and/or information services offered by the franchise (e.g., entertainment channel lineup, data packages, etc.) may differ from system to system. Some large cable companies operate several cable communication systems (e.g., in some cases up to hundreds of systems), and are known generally as Multiple System Operators (MSOs).
Cable Communication System Overview
FIG. 1 generally illustrates various elements of a conventional hybrid fiber-coaxial (HFC) cable communication system 160. The cable communication system 160 includes a headend 162 coupled to one or more nodes 164A, 164B and 164C via one or more physical communication media. The physical communication media typically include fiber optic cable and coaxial cable to convey information (e.g., television programming, Internet data, voice services) between the headend 162 and subscriber premises served by the nodes 164A, 164B and 164C of the cable communication system 160.
In FIG. 1, a first node 164A is illustrated with some detail to show multiple subscriber premises 190 as well as additional elements that similarly may be found in the other nodes 164B and 164C. In general, the headend 162 transmits information to and receives information from a given node via physical communication media (i.e., fiber optic cable and coaxial cable) dedicated to serving the geographic area covered by the node. Although the physical communication media of a given node may pass proximate to several premises, not all premises passed are necessarily subscriber premises 190 (i.e., actual subscribers to the services provided by the cable communication system 160); in some conventional cable communication systems, subscriber premises 190 of a given node may constitute on the order of 50% of the total number of premises passed by the physical communication media serving the node.
Although FIG. 1 illustrates only three subscriber premises 190 in the first node 164A, it should be appreciated that the geographic area covered by a representative node of a conventional cable communication system typically includes anywhere from approximately 100 premises to as many as 1000 premises (not all of which may be subscriber premises 190). Also, while FIG. 1 shows only three nodes 164A, 164B and 164C coupled to the headend 162, it should be appreciated that cable communication systems similar to the system 160 shown in FIG. 1 may include different numbers of nodes (e.g., for some larger cable communication systems, the headend may serve several hundreds of nodes).
Nodes
The first node 164A shown in FIG. 1 is depicted generally as either a “Fiber to the Neighborhood” (FTTN) node (also sometimes referred to as a “Fiber to the Feeder” or FTTF node), or a “Fiber to the Curb” (FTTC) node. In an FTTN/FTTF or FTTC node, fiber optic cable is employed as the physical communication medium to communicate information between the headend 162 and the general geographic area of subscriber premises. Within the area occupied by the subscriber premises, coaxial cable is employed as the physical communication medium between the fiber optic cable and respective subscriber premises 190. A general difference between FTTN/FTTF and FTTC nodes relates to how close the fiber optic cable comes to the premises in the node, and how many premises are passed by the coaxial cable portion of the node; for example, in an FTTC node, the fiber optic cable generally comes closer to the premises in the node than in an FTTN/FTTF node, and the coaxial cable portion of the FTTC node typically passes fewer than 150 premises (whereas the coaxial cable portion of an FTTN/FTTF node passes as many as from 200 to 1000 premises, as discussed further below). Unlike cable communication systems employing FTTN/FTTF and FTTC nodes, “Fiber to the Home” (FTTH) systems (also knows as “Fiber to the Premises” or FTTP systems) have a primarily fiber optic cable infrastructure (a “passive optical network” or PON) that runs directly and respectively to some smaller number of subscriber premises (e.g., approximately 30 or fewer premises passed).
As shown in FIG. 1, the first node 164A has an infrastructure (also referred to generally herein as a “cable plant”) that includes a first fiber optic cable 163A, a first optical/radio frequency (RF) converter 167, a first RF hardline coaxial cable plant 180, a plurality of first subscriber service drops 163C, and a plurality of first subscriber premises 190.
More specifically, the first node 164A includes a first fiber optic cable 163A, coupled to the headend 162 of the cable communication system 160 and to a first optical/radio frequency (RF) converter 167 (also sometimes referred to as a “bridge converter”) within the first node 164A. As noted above, depending on the configuration of the node as an FTTN/FTTF node or an FTTC node, the first optical/RF bridge converter 167 may be physically disposed at various geographic locations covered by the first node 164A. The bridge converter 167 generally serves to convert optical signals transmitted by the headend 162 to radio frequency (RF) signals that are received by subscriber premises 190 in the first node; the bridge converter 167 also converts RF signals transmitted by the subscriber premises 190 to optical signals that are received at the headend 162.
The first node 164A also includes a first RF hardline coaxial cable plant 180 (also referred to herein simply as a “hardline cable plant”) coupled to the bridge converter 167. The first hardline cable plant 180 constitutes another portion of the physical communication media over which information is carried, in the form of RF signals (e.g., modulated RF carrier waves), between the optical/RF bridge converter 167 and the subscriber premises 190 of the first node. Additional details of the first hardline cable plant 180 are discussed below in connection with FIG. 2.
As shown in FIG. 1, the first node 164A further includes multiple first subscriber service drops 163C, coupled to the first hardline cable plant 180 and respectively associated with subscriber premises 190. Each of the subscriber premises 190 includes one or more end-user modems 165 (also referred to herein as “subscriber modems” or “media terminal adapters”) to demodulate RF signals carrying data and/or voice information and received from the first hardline plant 180 via the premises' corresponding subscriber service drop 163C (a different device, commonly known as a “set-top box,” is typically employed at a subscriber premises to demodulate RF signals carrying video information). The subscriber modem(s) 165 also modulate an RF carrier with information (e.g., data and/or voice information) to be transmitted from the subscriber premises 190 to the first hardline cable plant 180. Thus, the first subscriber service drops 163C communicatively couple the subscriber modem(s) 165 of the respective subscriber premises 190 to the first hardline cable plant 180.
In the cable communication system 160 of FIG. 1, the first cable hardline plant 180 (as well as the first subscriber service drops 163C) carries RF signals that convey downstream information 183 from the headend 162 (as received via the fiber optic cable 163A and the bridge converter 167) to the subscriber premises 190 of the first node 164A. The first hardline cable plant also carries RF signals that convey upstream information 184 from at least some of the subscriber premises 190 of the first node 164A to the bridge converter 167 (which upstream information ultimately is transmitted to the headend 162 via the fiber optic cable 163A). To this end, the RF communication bandwidth supported by the first hardline cable plant 180 typically is divided into a downstream path bandwidth 181 in which the downstream information 183 is conveyed, and an upstream path bandwidth 182 in which the upstream information 184 is conveyed. In most conventional cable communication systems in the United States, the upstream path bandwidth 182 includes a first frequency range of from 5 MHz to 42 MHz (in other geographies, the upstream path bandwidth may extend to a higher frequency; for example, in Europe the upstream path bandwidth includes frequencies from 5 MHz to 65 MHz). The downstream path bandwidth 181 includes a second frequency range of from 50 MHz to 750 MHz (and in some instances as high as approximately 1 GHz). The downstream information 183 is conveyed by one or more downstream RF signals having a carrier frequency falling within the downstream path bandwidth 181, and the upstream information 184 is conveyed by one or more upstream RF signals having a carrier frequency falling within the upstream path bandwidth 182.
As noted above, the nodes 164B and 164C typically cover different geographic areas within the overall operating footprint of the cable communication system 160, but may be configured similarly to the first node 164A with respect to the various infrastructure constituting the node (e.g., each of the nodes 164B and 164C may include a dedicated fiber optic cable, optical/RF bridge converter, hardline plant, subscriber premises, and subscriber service drops to subscriber premises).
As also noted above, the overall infrastructure of a given node is referred to generally herein as a “cable plant,” with respective constituent elements of the cable plant including the first fiber optic cable 163A, the first optical/radio frequency (RF) converter 167, the first RF hardline coaxial cable plant 180, the plurality of first subscriber service drops 163C, and the plurality of first subscriber premises 190, as illustrated in FIG. 1. These respective elements have corresponding roles and functions within the cable plant (and the cable communication system as a whole); accordingly, it should be appreciated that while “cable plant” may refer to any one or more node infrastructure elements in combination, specific elements of the cable plant are referred to with particularity when describing their corresponding roles and functions in the context of the inventive concepts discussed in subsequent sections of this disclosure. For example, “RF hardline coaxial cable plant” (or “hardline cable plant”) refers specifically to the element 180 as introduced above in connection with FIG. 1, described further below in connection with FIG. 2, and similarly implemented according to various embodiments of inventive concepts discussed in subsequent sections of this disclosure.
In particular, FIG. 2 illustrates additional details of the first hardline cable plant 180 of the first node 164A. FIG. 2 also shows the first optical/RF converter 167 of the first node (to which the first hardline cable plant 180 is coupled), as well as one subscriber premises 190 of the first node (coupled to the first hardline cable plant 180 via a subscriber service drop 163C). Although only one subscriber premises 190 is shown in FIG. 2 for purposes of illustration, it should be appreciated that multiple subscriber premises may be coupled to the hardline cable plant 180 (e.g., as shown in FIG. 1). In FIG. 2, the first hardline cable plant 180 is indicated generally with dashed lines so as to distinguish various elements of the hardline cable plant 180 from the optical/RF converter 167 and other elements of the cable communication system generally associated with one or more subscriber premises 190. As noted above, hardline cable plants employed in other nodes of the communication system 160 shown in FIG. 1 generally may include one or more of the various elements shown in FIG. 2 as constituting the first hardline cable plant 180, and may be similarly configured to the first hardline cable plant 180.
As conventional cable communication systems have evolved over the years, so has some nomenclature for various elements of the system and, particularly, the hardline cable plant. Turning again to FIG. 2, a first segment of the hardline coaxial cable 163B in the hardline cable plant 180, between the optical/RF bridge converter 167 and a first amplifier 187 (e.g., in which power supply 186 is connected via connector 193), is sometimes referred to as an “express feeder” (historically, an express feeder was sometimes considered/referred to as part of the “trunk”). An express feeder may run for various distances and generally does not include any distribution taps 188. Conversely, a section of the hardline cable plant including one or more segments of hardline coaxial cable 163B and one or more distribution taps 188 sometimes is referred to merely as a “feeder” (as opposed to an “express feeder”). It should be appreciated that the terminology “trunk,” “express feeder,” and “feeder” are merely referred to above as examples of nomenclature used in the industry for various portions of the cable communication system and hardline cable plant. In exemplary implementations, various elements of the hardline cable plant 180 often are disposed above the ground, e.g., mounted on and/or hung between utility poles, and in some cases elements of the hardline cable plant also or alternatively may be buried underground.
As shown in FIG. 2, the first hardline cable plant 180 includes one or more segments of hardline coaxial cable 163B (one of which segments is coupled to the optical/RF converter 167). The hardline cable plant 180 also may include one or more components generally categorized as an “active” component, a “passive” component, a power supply, a connector, or various hardware (e.g., clamps, hangers, anchors, lashing wire, etc.) employed to secure various components to each other or other supporting infrastructure (e.g., utility poles, underground conduit, etc.). More specifically, with reference to FIG. 2, the hardline cable plant may include: one or more amplifiers 187 (also sometimes referred to as “line extenders”) constituting an active component and requiring power from one or more power supplies 186; one or more passive components, examples of which include distribution taps 188 (also referred to simply as “taps”), directional couplers 189 (also referred to as “splitters” or “combiners”), line terminators 191, and filters/attenuators (not shown explicitly in FIG. 2, although a filter/attenuator may be a constituent component of a tap, splitter/combiner, or a line terminator); one or more connectors or “fittings” 193 for coupling segments of the hardline coaxial cable 163B to various other elements of the hardline cable plant 180 (e.g., pin-type connectors, such as housing terminators, extension fittings, 90-degree fittings, splice connectors, etc., or one or more “splice blocks” 195 that may be employed to interconnect two segments of hardline coaxial cable 163B). FIGS. 3A through 3G illustrates examples of these various elements, which are discussed in greater detail in turn below.
With respect to the hardline coaxial cable 163B used in the hardline cable plant 180, as shown in FIG. 3A the coaxial cable commonly employed in the hardline plant often includes a center solid conductor surrounded by an electrically insulating material and a solid conductor shield to provide for improved electrical characteristics (e.g., lower RF signal loss/leakage) and/or some degree of environmental robustness. Some types of coaxial cables used for the hardline plant 180 include low density foam (LDF) insulation, which has insulating qualities similar to dry air, making it particularly well-suited for outdoor use. The solid conductor shield generally makes the cable somewhat more difficult to bend (hence the terminology “hardline” coaxial cable). In various implementations, 0.75 inch hardline coaxial cable may be employed for “express feeders,” whereas 0.625 inch hardline coaxial cable may be employed for “feeders.” One example of hardline coaxial cable 163B conventionally employed in the hardline plant 180 is given Commscope PIII 0.625 cable (e.g., see http://www.commscope.com/broadband/eng/product/cable/coaxial/1175378—7804.html). However, it should be appreciated that a variety of hardline coaxial cables may be employed in different hardline plants and/or different portions of the same hardline plant. Additionally, hardline tri-axial cable also is available that includes an additional shield layer to discourage electromagnetic interference, and may in some instances be employed in a hardline plant (for purposes of the present disclosure, any reference to “hardline coaxial cable” should be understood to include hardline tri-axial cable as well).
With reference again to FIG. 2, as noted above the hardline cable plant 180 also may include one or more power supplies 186 and one or more amplifiers 187 or “line extenders” (also shown in FIG. 3F). An exemplary power supply 186 converts commercially-available power (e.g., 120 Volts A.C. rms, 60 Hz) to voltage amplitudes (e.g., 60 VAC, 90 VAC) that may be distributed (e.g., in some cases along with RF signals via the hardline coaxial cable 163B) for providing power to one or more amplifiers 187 or other active components of the hardline cable plant. One or more amplifiers 187 may be employed to boost attenuated RF signals for further propagation or distribution along the hardline cable plant 180 (in one or both of the upstream path bandwidth or the downstream path bandwidth). Some types of amplifiers 187 may be bi-directional and provide separate amplification pathways for downstream and upstream RF signals, respectively. It should be appreciated that for purposes of the present discussion, the term “amplifier” is used generally to refer to a device that may amplify a signal; in some examples, an amplifier also may implement a filtering function as well (e.g., selective attenuation/amplification at one or more particular frequencies or over one or more frequency bands) for one or more RF signals propagating along the hardline cable plant 180. In particular, hardline cable plant amplifiers 187 typically include “diplex filters” that allow passage of signals through the amplifier only in the frequency ranges prescribed for the upstream path bandwidth and the downstream path bandwidth, respectively.
In conventional implementations of hardline coaxial cable plants, amplifiers may be distributed along the hardline coaxial cable plant of a given node at distances of approximately 1200 feet between amplifiers. One typical characterization of a node is referred to as “cascade,” which refers to the number of amplifiers in the longest branch of the hardline coaxial cable plant in the node. More specifically, the cascade for a given node often is denoted as “NODE+N,” in which N denotes the number of amplifiers between the RF/optical bridge converter of the node and an endpoint of the longest branch of the hardline coaxial cable plant in the node. With reference to FIG. 2, the illustrated example of the hardline cable plant 180 includes two amplifiers 167; if this illustration represented the entire hardline cable plant in the first node 164A, the cascade for this node would be referred to as “NODE+2.” In many conventional implementations of cable communication systems, typical cascades for hardline coaxial cable plants in respective nodes of the system are five or six (i.e., NODE+5 and NODE+6) (see section 3.1, pages 3-4 of “Architecting the DOCSIS Network to Offer Symmetric 1 Gbps Service Over the Next Two Decades,” Ayham Al-Banna, The NCTA 2012 Spring Technical Forum Proceedings, May 21, 2012, hereafter “Al-Banna,” which publication is hereby incorporated herein by reference in its entirety).
The hardline cable plant of FIG. 2 also may include one or more directional couplers 189 (also shown in FIG. 3E) to divide an input RF signal into two or more RF output signals or combine multiple input RF signals into one RF output signal (directional couplers also are referred to as “splitters” or “combiners”). For example, a splitter may divide an RF signal on one feeder section of the hardline cable plant to provide respective RF signals on two different feeder sections of the hardline cable plant; conversely, a directional coupler acting as a combiner combines RF signals from respective different feeders onto a same feeder of the hardline plant. In some examples, a directional coupler may include a transformer to split or combine power while maintaining a certain impedance. In other examples, a directional coupler 189 may include various features and materials to reduce interference. In common implementations, directional couplers are bi-directional devices in which both upstream RF signals and downstream RF signals may be present, wherein the directional coupler acts as a splitter with respect to downstream RF signals and a combiner with respect to upstream RF signals as these signals propagate along different feeder sections of the hardline cable plant.
A distribution tap (or simply “tap”) 188 of the hardline cable plant (see FIG. 3G) provides a connection point between the hardline cable plant and a subscriber service drop 163C. In one aspect, a tap functions similarly to a directional coupler in that a small portion of one or more downstream RF signals on the hardline coaxial cable 163B (e.g., in a “feeder” of the hardline plant) is extracted for providing to a subscriber premises 190. In the upstream direction, taps may be configured with different predetermined attenuation values (e.g., 4 dB, 11 dB, 17 dB, 20 dB) for attenuating RF signals originating from a subscriber premises 190 (e.g., signals transmitted by the subscriber modem 165) and intended for propagation along the hardline cable plant 180 toward the headend 162 of the cable communication system 160. Taps 188 may come in various forms, including multi-port taps. Taps typically include threaded connector ports to facilitate coupling to one or more hardline coaxial cable(s) and one or more subscriber service drops. In common examples, a port on a tap to which a subscriber service drop 163C is coupled may be constituted by a female F-type connector or jack, and the subscriber service drop 163C includes a coaxial cable terminated with a male F-type connector for coupling to the port of the tap 188 (e.g., see FIG. 11 and the discussion below in connection with same relating to male connector 197A and female connector 197B). Thus, in one aspect, the female F-type connector(s) of one or more taps 188 of the hardline cable plant 180 serve as a “boundary” between the hardline cable plant and other elements of the cable communication system generally associated with one or more subscriber premises 190.
Line terminators 191 of the hardline cable plant 180 (see FIG. 3C) electrically terminate RF signals at the end of a feeder to prevent signal interference. Line terminators 191 may include various materials and provide differing levels of shielding from environmental elements.
Various connectors 193 (see FIG. 3B) employed in the hardline cable plant 180, also referred to herein as “fittings,” may join two coaxial cables from separate sheaths, or may join a coaxial cable to one of the elements discussed above (e.g., amplifiers, power supplies, taps, directional couplers, line terminators, etc.). Connectors may be male, female, or sexless; some connectors have female structures with slotted fingers that introduce a small inductance; other connectors involve pin-based structures (e.g., pin-type connectors, such as housing terminators, extension fittings, 90-degree fittings, splice connectors, etc.). One common example of a connector is given by “F” series connectors, which may have ⅜-32 coupling thread or may be push-on. Other types of connectors employed in hardline cable plants include UHF connectors, BNC connectors, and TNC connectors. Various connectors differ in the methods they use for connecting and tightening. A splice block 195 (see FIG. 3D) is a particular type of connector used to join two respective segments of hardline coaxial cable.
As also shown in FIG. 2, the subscriber service drop 163C generally refers to the coaxial cable and associated hardware between a distribution tap 188 and a subscriber premises 190. In one aspect, as discussed above, a subscriber service drop 163C may be deemed to “begin” at a male F-connector (coupled to a female F-connector of a distribution tap 188) with which a coaxial cable used for the subscriber service drop 163C is terminated (e.g., see FIG. 11, male connector 197A). A subscriber service drop 163C often is constituted by a coaxial cable segment of a different type than the hardline coaxial cable 163B employed in the hardline plant 180 (as generally shorter cable lengths, greater physical flexibility, and less environmental robustness are required for subscriber service drops 163C than for the hardline cable plant 180; also whereas hardline coaxial cable is intended to be an essentially permanent component over the life of a cable communication system, subscriber service drops are considered as less permanent and may be installed and removed based on service changes relating to new subscribers or cancellation of services by existing subscribers). Some examples of coaxial cable conventionally employed for subscriber service drops 163C are given by RG-6 and RG-59 cables (e.g., see http://www.tonercable.com/assets/images/ProductFiles/1830/PDFFile/TFC %20T10%2059%20Series %20Drop %20Cable.pdf). In other examples, a subscriber service drop 163C may be constituted by a “flooded” cable or a “messenger” (aerial) cable; “flooded” cables may be infused with heavy waterproofing for use in an underground conduit or directly buried in the ground, whereas “messenger” cables may contain some waterproofing as well as a steel messenger wire along the length of the cable (to carry tension involved in an aerial drop from a utility pole). At the subscriber premises 190, the service drop 163C typically is fastened in some manner to the subscriber premises 190 and coupled to a ground block 198, and in turn connects to various components inside the subscriber premises, such as interior cables 192 (each of which typically terminates with connectors 196), one or more splitters/combiners 194, and one or more end user modems 165 (sometimes collectively referred to as “subscriber premises equipment” or “customer premises equipment”).
Finally, FIG. 2 also illustrates that an analyzer 110 (e.g., a spectrum analyzer and/or a tuned receiver) may be coupled to a junction between the bridge converter 167 and the hardline cable plant 180 so as to monitor RF signals that are transmitted to and/or received from the first node 164A. The coupling of the analyzer 110 to the junction between the bridge converter 167 and the hardline cable plant 180 is shown in FIG. 2 using dashed lines, so as to indicate that the analyzer 110 is not necessarily included as a constituent element of the first node, but may be optionally employed from time to time as a test instrument to provide information relating to signals propagating to and/or from the first node. As discussed further below in connection with FIGS. 1 and 4, an analyzer similarly may be employed in the headend to monitor various RF signals of interest in the cable communication system.
Table 1 below provides some typical parameters generally representative of node architecture found in several conventional communication systems (e.g., see “Mission is Possible: An Evolutionary Approach to Gigabit-Class DOCSIS,” John Chapman et al., The NCTA 2012 Spring Technical Forum Proceedings, May 21, 2012, hereafter referred to as “Chapman,” which publication is hereby incorporated herein by reference in its entirety; pages 35-47 and 57-62 of Chapman discuss particulars of node architecture):
TABLE 1Households Passed (HHP)500Subscriber Premises (e.g., high speed data)50%HHP Density75 HHP/mileNode Mileage6.67 milesCascadeNODE + 5 or +6Amplifiers/Mile4.5/mileTaps/Mile30/mileAmplifiers 30Taps200Highest Tap Value23 dBLowest Tap Value8 dBExpress Feeder Cable Type0.750 inch PIIILargest Express Feeder Span2000 feetFeeder (distribution) Cable Type0.625 inch PIIIFeeder Cable Distance to First Tap100 feetLargest Feeder Span1000 feetSubscriber Drop Cable TypeSeries 6Largest Drop Cable Span150 feetMaximum Subscriber Modem Transmit Power65 dBmV
Headend
With reference again to FIG. 1, the headend 162 of the cable communication system 160 generally serves as a receiving and processing station at which various entertainment program signals (e.g., television and video programming from satellite or land-based sources) are collected for retransmission to the subscriber premises of respective nodes 164A, 164B, and 164C over the downstream path bandwidth of each node. The headend 162 also may serve as a connection point to various voice-based services and/or Internet-based services (e.g., data services) that may be provided to the subscriber premises of respective nodes 164A, 164B, and 164C; such voice-based services and/or Internet-based services may employ both the upstream path bandwidth and downstream path bandwidth of each node. Accordingly, the headend 162 may include various electronic equipment for receiving entertainment programming signals (e.g., via one or more antennas and/or satellite dishes, tuners/receivers, amplifiers, filters, etc.), processing and/or routing voice-related information, and/or enabling Internet connectivity, as well as various electronic equipment for facilitating transmission of downstream information to, and receiving upstream information from, the respective nodes. Some conventional cable communication systems also include one or more “hubs” (not shown in FIG. 1), which are similar to a headend, but generally smaller in size; in some cable communication systems, a hub may communicate with a larger headend, and in turn provide television/video/voice/Internet-related services only to some subset of nodes (e.g., as few as a dozen nodes) in the cable communication system.
Since each node of the cable communication system 160 functions similarly, some of the salient structural elements and functionality of the headend 162 may be readily understood in the context of a single node (e.g., represented in FIG. 1 by the first node 164A). Accordingly, it should be appreciated that the discussion below regarding certain elements of the headend 162 particularly associated with the first node 164A applies similarly to other elements of the headend that may be associated with and/or coupled to other nodes of the cable communication system 160.
As shown in FIG. 1, the fiber optic cable 163A of the first node 164A is coupled to an optical/RF bridge converter 175 within the headend 162 (also referred to herein as a “headend optical/RF bridge converter”). As also shown in FIG. 1, each of the other nodes 164B and 164C similarly is coupled to a corresponding optical/RF bridge converter of the headend 162. The headend bridge converter 175 functions similarly to the bridge converter 167 of the first node; i.e., the headend bridge converter 175 converts upstream optical signals carried by the fiber optic cable 163A to RF signals 177 within the headend 162. In some implementations, the headend bridge converter 175 is constituted by two distinct devices, e.g., a downstream transmitter to convert RF signals originating in the headend to downstream optical signals, and an upstream receiver to convert upstream optical signals to RF signals in the headend. The headend 162 also may include an RF splitter 173, coupled to the headend bridge converter 175, to provide multiple paths (e.g., via multiple ports of the RF splitter) for the RF signals 177 in the headend that are transmitted to or received from the headend bridge converter 175. As discussed in greater detail below in connection with FIG. 4, the RF splitter 173 provides for various equipment (e.g., demodulators, modulators, controllers, test and monitoring equipment) to be coupled to the RF signals 177 within the headend carrying information to or from the first node 164A; for example, FIG. 1 illustrates an analyzer 110 (e.g., a spectrum analyzer), coupled to the RF splitter 173, that may be employed to monitor RF signals 177 in the headend 162 that are transmitted to and/or received from the first node 164A (as also discussed above in connection with FIG. 2).
The headend 162 shown in FIG. 1 also includes a cable modem termination system (CMTS) 170 that serves as the central controller for the subscriber modems in respective nodes of the cable communication system 160. In general, the CMTS 170 provides a bridge between the cable communication system 160 and an Internet Protocol (IP) network and serves as an arbiter of subscriber time sharing (e.g., of upstream path bandwidth in each node) for data services. In particular, for upstream information transmitted from subscriber modems in a given node to the headend 162 (e.g., the upstream information 184 from the first node 164A), in example implementations the CMTS 170 instructs a given subscriber modem in a given node when to transmit RF signals (e.g., onto a corresponding subscriber service drop and the hardline plant of the given node) and what RF carrier frequency to use in the upstream path bandwidth of the node (e.g., the upstream path bandwidth 182 of the first node 164A). The CMTS 170 then demodulates received upstream RF signals (e.g., the RF signals 177 from the first node 164A) to recover the upstream information carried by the signals, converts at least some of the recovered upstream information to “outgoing” IP data packets 159, and directs the outgoing IP data packets to switching and/or routing equipment (not shown in FIG. 1) for transmission on the Internet, for example. Conversely, the CMTS 170 also receives “incoming” IP data packets 159 from the Internet via the switching and/or routing equipment, modulates RF carrier waves with data contained in the received incoming IP data packets, and transmits these modulated RF carrier waves (e.g. as RF signals 177) to provide at least some of the downstream information (e.g., the downstream information 183 of the first node 164A) to one or more subscriber modems in one or more nodes of the cable communication system.
As also indicated in FIG. 1, in some implementations in which the recovered upstream information includes voice information (e.g., from subscriber premises receiving VoIP services), the CMTS 170 may also direct “outgoing” voice information 157 to a voice switch coupled to a Public Switched Telephone Network (PSTN). The CMTS 170 also may receive “incoming” voice information 157 from the PSTN, and modulate the received incoming voice information onto RF carrier waves to provide a portion of the downstream information.
As illustrated in FIG. 1, the CMTS 170 may include multiple RF ports 169 and 171, in which typically one pair of RF ports 169 and 171 of the CMTS facilitates coupling of a corresponding node of the cable communication system 160 (in some instances via one or more RF splitters 173) to the CMTS 170; in particular, for the first node 164A shown in FIG. 1, downstream RF port 169 provides downstream information from the CMTS to the first node, and upstream RF port 171 provides upstream information to the CMTS from the first node. For each downstream RF port 169, the CMTS further includes one or more modulation tuners 172 coupled to the downstream RF port; similarly, for each upstream RF port 171, the CMTS includes one or more demodulation tuners 174 coupled to the upstream RF port 171. As noted above, the modulation tuner(s) 172 is/are configured to generate one or more modulated RF carrier waves to provide downstream information to subscriber modems of the node coupled to the corresponding RF port 169; conversely, the demodulation tuner(s) 174 is/are configured to demodulate one or more received upstream RF signals carrying upstream information from the subscriber modems of the node coupled to the corresponding RF port 171.
FIG. 4 illustrates further details of a portion of the headend 162 shown in FIG. 1, relating particularly to upstream information received from subscriber modems of the first node 164A via the fiber optic cable 163A, and exemplary arrangements of the CMTS 170. For example, FIG. 4 shows that the RF splitter 173 associated with the first node 164A may include multiple ports to couple upstream RF signals 177 received from the first node to each of the analyzer 110, one RF port 171 of the CMTS 170, a digital account controller 254 (DAC), and other test and/or monitoring equipment 256. The DAC 254 relates primarily to video programming (e.g., managing on-demand video services by receiving programming requests from subscriber premises “set-top boxes” and coordinating delivery of requested programming). As discussed elsewhere herein, the analyzer 110 may be configured to monitor a spectrum of the upstream path bandwidth to measure an overall condition of the upstream path bandwidth (e.g., a presence of noise in the node) and/or provide performance metrics relating to the conveyance of upstream information in the node (e.g., for diagnostic purposes). Other test and/or monitoring equipment 256 may be configured to receive signals from field-deployed monitoring devices (most typically in power supplies in the node) to alert system operators of critical events (e.g., a power outage) or other alarm conditions.
The CMTS 170 itself may be constructed and arranged as a modular apparatus that may be flexibly expanded (or reduced in size) depending in part on the number of nodes/subscribers to be served by the cable communication system 160. For example, the CMTS 170 may have a housing configured as a chassis with multiple slots to accommodate “rack-mountable” modular components, and various RF modulation/demodulation components of the CMTS may be configured as one or more such modular components, commonly referred to as “blades,” which fit into respective slots of the CMTS's chassis. FIG. 4 shows a portion of the CMTS 170 including two such “blades” 252.
As illustrated in FIG. 4, each blade 252 of the CMTS 170 may include multiple upstream RF ports 171 (e.g., four to six ports per blade), as well as one or more downstream ports (not explicitly shown in FIG. 4). Historically, each upstream RF port 171 of a blade 252 was coupled to only one demodulation tuner 174 serving a particular node coupled to the upstream RF port 171; in more recent CMTS configurations, a blade 252 may be configured such that one or more upstream RF ports 171 of the blade may be coupled to multiple demodulation tuners 174 (e.g., FIG. 4 shows two demodulation tuners 174 coupled to one upstream port 171 of the top-most blade 252). In this manner, the upstream information from a given node may be received by the CMTS via multiple RF signals 177 (i.e., one RF signal per demodulation tuner 174 coupled to the blade's upstream RF port 171 corresponding to the given node). The CMTS 170 may include virtually any number of blades 252, based at least in part on the number of nodes included in the cable communication system 160 (and the number of RF ports per blade).
Various implementations of the CMTS 170 constitute examples of a “cable modem system,” which generally refers to one or more modulation tuners and/or demodulation tuners, and associated controllers and other equipment as may be required, to facilitate communication of downstream information to, and/or upstream information from, one or more subscriber premises. As noted above, one or both of the downstream information and upstream information handled by a cable modem system may include a variety of data content, including Internet-related data, voice-related data, and/or audio/video-related data. Other implementations of a cable modem system may include a “Converged Cable Access Platform” (CCAP), which combines some of the functionality of a CMTS discussed above and video content delivery in contemplation of conventional MPEG-based video delivery migrating to Internet Protocol (IP) video transport (e.g., see “CCAP 101: Guide to Understanding the Converged Cable Access Platform,” Motorola whitepaper, February 2012, http://www.motorola.com/staticfiles/Video-Solutions/Products/Video-Infrastructure/Distribution/EDGE-QAM/APEX-3000/_Documents/_StaticFiles/12.02.17-Motorola-CCAP %20101_white %20paper-US-EN.pdf, which whitepaper is hereby incorporated by reference herein in its entirety). For purposes of the discussion below, the CMTS 170 is referred to as a representative example of a “cable modem system;” however, it should be appreciated that the various concepts discussed below generally are applicable to other examples of cable modem systems, such as a CCAP.
Communication Concepts
With reference again to FIG. 1, the transmission of downstream information 183 between the headend 162 and subscriber premises 190 in the first node 164A, and the transmission of upstream information 184 from one or more subscriber premises 190 in the first node 164A to the headend 162, may be understood as follows.
With respect to downstream information 183 in the first node 164A, digital information (e.g., voice information or other data in the form of IP data packets 159 from an external IP network) constituting the downstream information is modulated onto an RF carrier wave (having a particular carrier frequency in the downstream path bandwidth 181) by a modulation tuner 172 in the CMTS 170 at the headend 162, to provide a downstream RF signal 177 via a port 171 of the CMTS 170. This downstream RF signal 177 is converted to a downstream optical signal by headend optical/RF converter 175 and transported via first fiber optic cable 163A to the first optical/RF converter 167 in the first node 164A, which converts the downstream optical signal back to an RF signal. This converted RF signal (carrying the downstream information 183) is then transported via the hardline cable plant 180 and subscriber service drops 163C to the respective subscriber modems 165 of the subscriber premises 190, each of which modems includes appropriate demodulator circuitry that is tuned to the carrier frequency of the downstream RF signal so as to appropriately demodulate the RF signal and thereby recover the downstream information 183 (e.g., in the form of the IP data packets 159).
With respect to upstream information 184, the foregoing process is essentially reversed; i.e., digital information originating from a given subscriber premises 190 (e.g., voice information or other data in the form of IP data packets) constituting at least a portion of the upstream information 184 is modulated onto an RF carrier wave (having a particular frequency in the upstream path bandwidth 182) by modulation circuitry in the subscriber modem 165 to provide an upstream RF signal. This upstream RF signal is transported via subscriber service drop 163C and the hardline cable plant 180 to the first optical/RF converter 167 in the first node 164A, which converts the upstream RF signal to an upstream optical signal that is transported to the headend 162 via the first fiber optic cable 163A. At the headend, the headend optical/RF converter 175 converts the upstream optical signal back to an upstream RF signal 177. This RF signal 177 is coupled via a port 171 of the CMTS 170 to a demodulation tuner 174 tuned to the carrier frequency of the upstream RF signal so as to appropriately demodulate the upstream RF signal and thereby recover the upstream information 184 (e.g., which may be in the form of IP data packets 159 to be conveyed to an external IP network).
Regarding common modulation schemes that may be employed generally by subscriber modems 165, or the modulation tuners 172 and demodulation tuners 174 of the CMTS 170 at the headend 162, to encode digital information (e.g., the downstream information 183 and the upstream information 184) on an RF carrier wave to provide an RF signal, such modulation schemes often employ modulation of the phase and/or the amplitude of a carrier wave having a given carrier frequency ƒ based on the particular digital information to be encoded. To illustrate some common digital modulation schemes, it is helpful to first represent a sinusoidal carrier wave having an amplitude A, a frequency f, and a phase φ (in radians), denoted mathematically as A sin (2πft+φ), as the composition of two sinusoidal waves that are out of phase by 90 degrees with respect to each other. Using well-known trigonometric relationships, it may be shown that:A sin(2πft+φ)=I sin(2πft)+Q cos(2πft),  Eq. 1where:
                    A        =                                            I              2                        +                          Q              2                                                          Eq        .                                  ⁢        2                        and                                                      φ        =                              arctan            ⁡                          (                              Q                I                            )                                .                                    Eq        .                                  ⁢        3            The foregoing decomposition of a carrier wave is sometime referred to as “orthogonal decomposition,” in which the sine term of the decomposition is referred to as the “in-phase” component having an amplitude I, and the cosine term of the decomposition is referred to as the “quadrature” component having an amplitude Q. The representation of a sinusoidal carrier wave in terms of in-phase and quadrature components may be facilitated by a coordinate plane defined by a horizontal axis representing values of I (the “in-phase component” axis) and a vertical axis representing values of Q (the “quadrature component” axis). FIG. 5 shows a generic example of the carrier wave of Eq. 1 represented as a vector on such a coordinate plane, in which the length of the vector is given by the amplitude A (according to Eq. 2) and the phase of the vector is given by the angle ca between the vector and the in-phase component axis (according to Eq. 3).
In some digital modulation schemes commonly employed in conventional cable communication systems, the amplitudes I and Q of the in-phase and quadrature components, respectively, may only have one of some number of finite values at any given time. Once the in-phase and quadrature components are combined according to Eq. 1, each of the possible combinations of finite values that the amplitudes I and Q may have according to a given digital modulation scheme correspond to a particular unique state of the resulting modulated carrier wave, which state is defined by a particular amplitude A and a particular phase φ of the resulting modulated carrier wave (pursuant to Eq. 2 and Eq. 3 above). With the foregoing in mind, each of the possible combinations of finite values that the amplitudes I and Q may have in a given modulation scheme are assigned to some number m bits of digital information to be encoded on the carrier wave; each m bits of digital information that is assigned to a particular combination of I and Q values is commonly referred to as a “symbol.” In this manner, each unique combination of m digital bits (each “symbol”), representing a particular combination of possible values for each of I and Q, “maps” to a particular amplitude A and a particular phase φ of the resulting modulated carrier wave.
More specifically, a given modulator (e.g., modulator circuitry in a subscriber modem 165, or a modulation tuner 172 of the CMTS 170) separates a carrier wave to be encoded into respective in-phase and quadrature components and, based on the respective values (i.e., logic 1 or logic 0) of m digital bits in a given symbol to be encoded, selects corresponding assigned values for the amplitudes I and Q respectively. The in-phase and quadrature components are then recombined and, as noted above, the resulting modulated carrier wave has a particular amplitude A and a particular phase φ corresponding to the particular symbol encoded on the wave. A given demodulator employing the same modulation scheme and receiving such a modulated carrier wave (a “signal”) is thus able to recover the particular symbol by determining the amplitude and phase of the received signal.
Two common digital modulation schemes employed in conventional cable communication systems and based on the foregoing concepts include quadrature phase shift keying (QPSK) and different orders of quadrature amplitude modulation (QAM). With reference again to Eq. 1 above, in QPSK each of the amplitudes I and Q for the respective in-phase and quadrature components has the same magnitude |X| and one of two possible non-zero values at any given time, namely I=+X or −X, and Q=+X or −X. A convenient way to visualize a QPSK modulation scheme (as well as other digital modulation schemes) is via a “constellation diagram,” in which different possible states of the recombined (modulated) carrier wave are illustrated in the coordinate plane employed for FIG. 5 (i.e., having a horizontal “in-phase” axis and a vertical “quadrature” axis). FIG. 6 illustrates such a constellation diagram 5000A for a QPSK modulation scheme; again, in QPSK, only values of +X and −X are possible for each of I and Q. As such, there are four different possible states for the resulting recombined carrier wave, which states commonly are referred to as “constellation points.” An interesting artifact of QPSK is that, since the magnitudes of I and Q are identical at any given time (i.e., |X|), the amplitude A of the resulting modulated carrier wave remains fixed according to Eq. 2 (i.e., A=√{square root over (2)}|X|); however, the respective constellation points have different phases φ, namely 45 degrees, 135 degrees, 225 degrees and 315 degrees (hence the name quadrature “phase shift” keying). Given the four constellation points representing different phases, each point may be represented by a unique combination of two bits of digital information constituting a “symbol” corresponding to the constellation point (e.g., “11”=45 degrees; “01”=135 degrees; “00”=225 degrees, and “10”=315 degrees).
Like QPSK, Quadrature Amplitude Modulation (QAM) similarly is based on amplitude modulation of respective in-phase and quadrature components of a carrier wave, which components are recombined to form an information-bearing signal (i.e., a modulated carrier wave). Unlike QPSK, however, in QAM each of the amplitudes I and Q of the respective in-phase and quadrature components may have one of multiple different magnitudes, resulting in a modulated carrier wave with both changing amplitude A and phase ca. Different QAM schemes may be visualized via a constellation diagram similar to that shown in FIG. 6 for QPSK. In QAM, the constellation points typically are arranged in a grid with equal vertical and horizontal spacing. The number of constellation points in a given square-grid QAM implementation is often referred to as the QAM “modulation order” and is related to the number n of unique magnitudes each of the I and Q amplitudes may have. If the value of n for the I amplitudes is different than the number n for Q amplitudes, a rectangular-shaped constellation diagram results; if on the other hand the number n of unique magnitudes is the same for both I and Q), a square constellation diagram results. Rectangular QAM constellations generally are sub-optimal in that constellation points are not maximally spaced for a given constellation energy, and they are somewhat more challenging to modulate and demodulate. Accordingly, square QAM constellations are more commonly (but not exclusively) employed in conventional cable communication systems, wherein the QAM modulation order is given by:QAM modulation order=4n2.  Eq. 4From Eq. 4, it may be appreciated that the QPSK constellation diagram shown in FIG. 6 is actually a special case of QAM with modulation order 4 (i.e., n=1 results in 4-QAM).
The number of constellation points in a given QAM implementation also dictates the number of unique symbols that may be mapped to the constellation diagram, which in turn depends on the number of bits per symbol m; i.e., the number of unique symbols=2^m (m=1, 2, 3 . . . ). Again, since QAM often is implemented as a square-grid constellation diagram (e.g., see Eq. 4), certain QAM modulation orders are more commonly implemented in conventional cable communication systems, for values of m>2 and integer values of n that satisfy:QAM modulation order=4n2=2^m.  Eq. 5
Table 2 below lists some common QAM modulation orders and associated values of n (number of unique magnitudes of I and Q) and m (number of bits per symbol) for square QAM constellations based on Eq. 5. Table 2 also includes entries for 32-QAM and 128-QAM and their associated bits per symbol m; these are rectangular constellations (for which there is no integer value for n in Eq. 5) that nonetheless may be employed in some cable communication system implementations. To illustrate the higher QAM modulation orders for square constellations listed in Table 2, FIG. 7 provides an example constellation diagram 5000B for 16-QAM, FIG. 8 provides an example constellation diagram 5000C for 64-QAM, and FIG. 9 provides an example constellation diagram 5000D for 256-QAM.
TABLE 2QAM modulation order(symbols per constellation)nm4121624325644612872568851291024161020481140963212
Although Table 2 and FIGS. 7 through 9 illustrate four exemplary modulation orders for QAM, it should be appreciated that pursuant to the general principles outlined above, a number of different QAM modulation orders are possible in addition to those noted in Table 2 and FIGS. 7 through 9. In general, by moving to a higher QAM modulation order it is possible to transmit more bits per symbol. However, for purposes of comparing two different QAM modulation orders, if the mean energy of the constellation is to remain the same, the respective constellation points must be closer together within the constellation. Recall that each constellation point represents a particular amplitude A and phase φ of a modulated carrier signal that ultimately needs to be demodulated by a demodulator (that can effectively discern amongst different points of the constellation). As discussed further below in connection with FIG. 16, noise that may be present on a physical communication medium carrying a QAM signal may alter one or both of the amplitude A and phase φ of the signal such that the signal, upon demodulation, may be confused with another neighboring point on the constellation, resulting in the wrong symbol being recovered by a demodulator. As respective constellation points are more “tightly-packed” in higher modulation order constellations, they are thus more susceptible to noise and other corruption upon demodulation; accordingly, higher modulation-order QAM can deliver more data less reliably than lower modulation-order QAM, for constant mean constellation energy.
Turning again to FIG. 1, and with respect to communication of upstream and downstream RF signals associated with the first node 164A within the headend 162, on the hardline plant 180, or on the subscriber service drops 163C, each of the downstream path bandwidth 181 and the upstream path bandwidth 182 is divided up into multiple communication “channels” to convey the downstream information 183 and the upstream information 184. For purposes of the present disclosure, a “physical communication channel” may be described by at least three parameters, namely: 1) a carrier frequency of an RF carrier wave onto which information (e.g., upstream information 184 or downstream information 183) is modulated; 2) a modulation type (e.g., QPSK, QAM), used by a modulation tuner 172 at the headend or modulation circuitry of a subscriber modem 165, to modulate the carrier wave; and 3) a channel bandwidth, wherein the carrier frequency typically is located at a center of the channel bandwidth. As discussed in detail further below, another parameter of an upstream physical communication channel may include the access protocol (e.g., Time Division Multiple Access, Advanced Time Division Multiple Access, Synchronous Code Division Multiple Access) employed to transport upstream information from multiple subscriber premises via the physical communication channel.
Regarding physical communication channel parameters, as noted above the upstream path bandwidth 182 in the United States typically includes upstream channels having carrier frequencies within a first frequency range of from 5 MHz to 42 MHz (5 MHz to 65 MHz in Europe), and the downstream path bandwidth 181 includes downstream channels having carrier frequencies within a second frequency range of from 50 MHz to 750 MHz (and in some instances as high as approximately 1 GHz). As discussed in greater detail below, practical considerations relating to noise have limited the information carrying capacity of the upstream path bandwidth in the portion of the spectrum between approximately 20 MHz and 42 MHz, and have rendered the portion of the upstream path bandwidth between 5 MHz and approximately 20 MHz effectively unusable. Accordingly, upstream channels having carrier frequencies in the range of 5 to approximately 20 MHz (and particularly below 18 MHz, and more particularly below 16.4 MHz, and more particularly below 10 MHz) are rarely if ever employed in conventional cable communication systems; if used at all, such channels are typically limited to rudimentary binary modulation schemes (e.g., binary phase-shift keyed or “PSK” modulation, or binary frequency-shift keyed or “FSK” modulation) rather than quadrature modulation schemes, and have significantly limited information-carrying capacity and functionality (e.g., conveying subscriber orders for pay-per-view television from subscriber premises to the headend of the cable communication system).
With respect to the bandwidth of physical communication channels, typical upstream and downstream channel bandwidths employed for cable communication system channels are 3.2 MHz and 6.4 MHz, although other channel bandwidths are possible (e.g., 1.6 MHz). Conceptually, the bandwidth of a physical communication channel for which QPSK or QAM modulation schemes are employed effectively represents the number of symbols per second that may be conveyed over the channel, which in turn relates to the maximum data rate of the channel. According to various channel filtering and tuning techniques which define the passband (i.e., shape or profile) of a channel as a function of frequency, only a portion of the stated bandwidth of a channel is available for data transmission (e.g., the lowest and highest frequency portions of the channel serve as transition bands and the centermost frequencies of the channel serve as a passband); thus, it is conventionally presumed that approximately 80% of the stated bandwidth of a given channel is deemed available for data transmission. Accordingly, the “symbol rate” of a given channel (i.e., the maximum number of symbols per second that can be effectively conveyed over the channel) is taken to be 80% of the channel's specified bandwidth. With the foregoing in mind, a maximum “deployed data rate” (also sometimes referred to as “raw data rate”) with which the upstream information 184 or the downstream information 183 may be conveyed on a given physical communication channel (“data rate” is also sometimes referred to as “channel capacity”) is typically specified in units of bits per second and is based on the symbol rate of the channel (0.8 BW) and the number m of bits per symbol (as dictated by the modulation order), given by:Data rate(bits/sec)=0.8 BW(symbols/sec)*m(bits/symbol).  Eq. 6Using values of m from Table 2 above corresponding to different QAM modulation orders (wherein 4-QAM=QPSK), FIG. 10 illustrates a bar graph showing different modulation orders and channel bandwidths along the horizontal axis, and corresponding maximum deployed (or “raw”) data rates along the vertical axis, according to Eq. 6. From FIG. 10, it may be seen that a conventional 6.4 MHz channel in which a 64-QAM modulation scheme is used may convey data at a maximum deployed data rate of approximately 30 Mbits/s.
As an alternative to the graph of FIG. 10, the maximum deployed data rates for respective QAM modulation orders may be normalized for different possible channel bandwidths, in units of bits/sec-Hz (by removing the BW term from Eq. 6). Table 3 below provides the normalized maximum “raw” data rates for different QAM modulation orders, and the corresponding maximum raw data rates (maximum deployed channel capacities) for channel bandwidths of 1.6 MHz, 3.2 MHz and 6.4 MHz, respectively, corresponding to each QAM modulation order:
TABLE 3QAM modulationRaw Data1.6 MHz BW3.2 MHz BW6.4 MHz BWorder (symbolsRate/HzRaw Data RateRaw Data RateRaw Data Rateper constellation)m(bps/Hz)(Mbits/s)(Mbits/s)(Mbits/s)QPSK (4-QAM)21.62.565.1210.24 16-QAM43.25.1210.2420.48 32-QAM54.06.4012.8025.60 64-QAM64.87.6815.3630.72 128-QAM75.68.9617.9235.84 256-QAM86.410.2420.4840.96 512-QAM97.211.5223.0446.081024-QAM108.012.8025.6051.202048-QAM118.814.0828.1656.324096-QAM129.615.3630.7261.44
With reference again to FIG. 1, it should be appreciated that in the conventional cable communication system 160, downstream information 183 in the first node 164A generally is broadcast from the headend 162, using multiple downstream channels having different carrier frequencies in the downstream path bandwidth 181, to all subscriber premises 190 in the node; however, the demodulator circuitry of a given subscriber modem 165 generally is tuned to only one or more particular carrier frequencies in the downstream path bandwidth 181 at a given time so as to recover only a particular portion of the downstream information 183 (i.e., particular downstream information encoded on an RF signal having a carrier frequency to which the modem's demodulator is tuned).
Conversely, multiple subscriber premises 190 in the node 164A typically share a single upstream channel defined by an RF signal having a carrier frequency in the upstream path bandwidth 182, so as to convey respective portions of upstream information 184 originating from different subscriber modems 165 that share the upstream channel. A collection of multiple subscriber premises/subscriber modems of a given node that share a single upstream channel commonly is referred to as a “service group” (in FIG. 1, such a service group is denoted by reference number 195; in some conventional cable communication systems, a service group may include between approximately 100 and 300 subscriber premises). To ensure that upstream information from multiple subscriber modems in a service group is effectively received at the headend, various upstream “access protocols” may be implemented by the CMTS 170 and the subscriber modems 165 to regulate the manner in which portions of upstream information from different subscriber modems are carried over the shared upstream channel. Examples of such access protocols include Time Division Multiple Access (TDMA), Asynchronous Transfer Mode (ATM), Carrier Sense Multiple Access/Collision Detection (CSMA/CD). Generally speaking, such access protocols are responsible for implementing timing schemes with which different subscriber modems may transmit portions (“transmission bursts”) of upstream information, and in some cases assigning a carrier frequency to be modulated (with upstream information) by the modulation circuitry of a subscriber modem.
One widely adopted specification for transport of upstream and downstream information via a cable communication system and associated access protocols is referred to as the “Data Over Cable Service Interface Specification” (DOCSIS). DOCSIS is an international open protocol developed by the industry consortium CableLabs for deploying high-speed data and voice services over cable communication systems similar to the system 160 shown in FIG. 1. The DOCSIS specification relates to aspects of the “physical layer” of the communication system (e.g., specifying channel bandwidths and modulation types supported), the “data link layer” of the communication system (e.g., specifying access protocols for transmission of upstream information, quality of service features to support Voice-over-Internet Protocol, or “VoIP,” and channel bonding), and the “network layer” of the communication system (e.g., management of subscriber modems and the CMTS via IP addresses).
More specifically, with respect to “physical layer” specifications, the North American version of DOCSIS utilizes 6 MHz channels for transmission of downstream information, and specifies upstream channel bandwidths of between 200 kHz and 3.2 MHz (DOCSIS version 1.0) and more recently 6.4 MHz (DOCSIS versions 2.0 and 3.0). All versions of DOCSIS specify that 64-QAM or 256-QAM may be used for modulation of downstream information; DOCSIS version 1.0 specified QPSK or 16-QAM for modulation of upstream information, and DOCSIS versions 2.0 and 3.0 specify QPSK, 8-QAM, 16-QAM, 32-QAM, and 64-QAM for modulation of upstream information (where noise conditions permit such higher modulation orders, as discussed further below). DOCSIS versions 2.0 and 3.0 also supports a limited special version of 128-QAM for modulation of upstream information, requiring trellis coded modulation in Synchronous Code Division Multiple Access (S-CDMA) mode (discussed further below). With respect to “data link layer” or media access control layer (MAC) specifications, DOCSIS employs a mixture of deterministic access methods for transmission of upstream transmission, specifically TDMA for DOCSIS version 1.0/1.1 and both ATDMA (Advanced Time Division Multiple Access) and S-CDMA for DOCSIS versions 2.0 and 3.0. For DOCSIS 1.1 and above the data link layer also includes quality-of-service (QoS) features to support applications that have specific traffic requirements such as low latency (e.g., VoIP, some gaming applications). DOCSIS version 3.0 also features channel bonding, which enables multiple downstream and upstream channels to be used together at the same time by a single subscriber modem.
DOCSIS also defines a “channel utilization index” which generally represents a percentage of time over some predetermined time period that the respective subscriber premises of a service group are transmitting upstream information and hence “using” the physical communication channel over which upstream information from the service group is conveyed. More specifically, the upstream channel utilization index is expressed as a percentage of minislots utilized on the physical communication channel, regardless of burst type. In one example, minislots are considered utilized if the CMTS receives an upstream burst from any subscriber modem transmitting on the physical channel. In another example (“contention REQ and REQ/DATA”), minislots for a transmission opportunity are considered utilized if the CMTS receives an upstream burst within the opportunity from any subscriber modem transmitting on the physical channel.
Egress and Ingress
A cable communication system is considered theoretically as a “closed” information transmission system, in that transmission of information between the headend 162 and subscriber modems 165 occurs via the physical communication media of optical fiber cable, a hardline cable plant, and subscriber service drops (and not over air or “wirelessly”) via prescribed portions of frequency spectrum (i.e., in the U.S., upstream path bandwidth from 5 MHz-42 MHz; downstream path bandwidth from 50 MHz to 750 MHz or higher). In practice, however, cable communication systems generally are not perfectly closed systems, and may be subject to signal leakage both out of and into the system (e.g., through faulty/damaged coaxial cable and/or other network components). The term “egress” refers to signal leakage out of a cable communication system, and the term “ingress” refers to signal leakage into a cable communication system. A significant operating and maintenance expense for owners/operators of cable communication systems relates to addressing the problems of signal egress and ingress.
More specifically, egress occurs when RF signals travelling in the downstream path bandwidth of a cable communication system leak out into the environment. Egress may cause RF interference with devices in the vicinity of the point of egress, and in some cases can result in weaker signals reaching the subscriber modems 165. The Federal Communications Commission (FCC) enforces laws established to regulate egress, noting that egress may cause interference with “safety-of-life” radio services (communications of police, fire, airplane pilots) and endanger the lives of the public by possibly hampering safety personnel's efforts. Accordingly, the FCC has set maximum individual signal leakage levels for cable communication systems. As a further prevention, the FCC requires cable communication system operators to have a periodic on-going program to inspect, locate and repair egress on their systems.
In light of the potential for catastrophic harm which may be caused by cable communication system egress interfering particularly with aeronautical navigational and communications radio systems, the FCC requires more stringent regulations for cable communication system egress in the aeronautical radio frequency bands (sometimes referred to as the “aviation band,” from approximately 110 MHz to 140 MHz). For example, any egress in the aviation band which produces a field strength of 20 uV/m or greater at a distance of three meters must be repaired in a reasonable period of time. Due to these regulations and government oversight by the FCC, cable communication system operators historically have focused primarily on egress monitoring and mitigation.
With respect to examples of conventional techniques for detecting egress, the company Comsonics, Inc. (http://www.comsonics.com/) provides various equipment (e.g., a GPS navigation system, an RF receiver, and an RF antenna), referred to as Genacis™, for a vehicle-based approach to monitor egress over a geographic region. In particular, the Genacis™ RF antenna monitors one or more particular frequencies in the downstream path bandwidth of the cable communication system generally corresponding to the aviation band (e.g., approximately 120 MHz) and records signal amplitude of any egress emanating from the cable communication system at the monitored frequency and vehicle position. This information is used to identify locations of egress in the network.
Ingress is noise or interference that may occur from an outside signal leaking into the cable communication system infrastructure. The source of the outside signal is commonly referred to as an “ingress source.” Some common ingress sources include broadband noise generated by various manmade sources, such as automobile ignitions, electric motors, neon signs, power-line switching transients, arc welders, power-switching devices such as electronic switches and thermostats, and home electrical appliances (e.g., mixers, can openers, vacuum cleaners, etc.) typically found at subscriber premises. Although some of these ingress sources produce noise events in the 60 Hz to 2 MHz range, their harmonics may show up in the cable communication system upstream path bandwidth from 5 MHz to 42 MHz. “Impulse” noise is generally characterized by a relatively short burst of broadband noise (e.g., 1 to 10 microseconds), and “burst” noise is generally characterized by bursts of broadband noise with durations up to about 100 microseconds. In addition to manmade sources of broadband noise which may contribute to burst or impulse noise, natural sources of burst noise include lightning and electrostatic discharge, which may give rise to noise events from 2 kHz up to 100 MHz.
Other ingress sources include relatively narrowband signals arising from transmission sources that may be proximate to the cable communication system (e.g., transmitting devices such as HAM or CB radios in the vicinity, subscriber premises garage door openers, fire and emergency communication devices, and pagers). In particular, ham radio operators use carrier frequencies at 7 MHz, 10 MHz, 14 MHz, 18 MHz, 21 MHz, 24 MHz and 28 MHz, and citizen band radios use frequencies at approximately 27 MHz, all of which fall within the upstream path bandwidth of the cable communication system.
The foregoing ingress sources often create intermittent and/or seemingly random signals that may leak into the infrastructure of the cable communication system, causing disturbances that may be difficult to locate and/or track over time. Such disturbances may impede normal operation of the cable communication system, and/or render some communication bandwidth significantly compromised or effectively unusable for conveying information. In particular, ingress from these random and/or intermittent sources may undesirably and unpredictably interfere with transmission of upstream information by operative RF signals in the upstream path bandwidth. Yet another ingress source includes “terrestrial” signals present in free space, primarily from short wave radio and radar stations (e.g., short wave radio signals are present from approximately 4.75 MHz to 10 MHz).
It is commonly presumed in the cable communication industry that egress may serve as a proxy for ingress; i.e., where there is an opening/fault in the cable communication system that allows for signal leakage from the system to the outside (egress), such an opening/fault likewise allows for outside signals to enter the cable communication system (ingress). It is also commonly presumed in the cable communication industry that a significant majority of cable communication system faults allowing for signal leakage into and out of the system occur almost entirely in connection with system elements associated with one or more subscriber premises; more specifically, subscriber service drops, and particularly subscriber premises equipment, are conventionally deemed to be the greatest source of signal leakage problems.
FIG. 11 shows the example subscriber premises 190 from FIG. 2, together with a portion of the hardline cable plant 180 including two segments of hardline coaxial cable 163B and a tap 188, to which the segments of hardline coaxial cable 163B are coupled via connectors 193. A subscriber service drop 163C also is coupled to the tap 188, for example, via a male connector 197A on one end of the subscriber service drop 163C and a female connector 197B of the tap 188. As illustrated in FIG. 11, it is commonly presumed in the cable communication industry that approximately 75% or more of ingress in a cable communication system originates inside respective subscriber premises 190 (all subscriber premises taken in aggregate), and that approximately 20% is attributable to the respective subscriber service drops 163C (taken in aggregate) (e.g., see “Return Path Maintenance Plan: A Five Step Approach to Ensuring a Reliable Communications Path,” Robert Flask, Acterna LLC whitepaper, 2005, page 7, http://sup.xenya.si/sup/info/jdsu/white_papers/ReturnPathMaintenancePlan_Whitepaper.pdf).
More specifically, poorly shielded subscriber premises equipment (e.g., defective or inferior quality cables 192; loose, corroded, or improperly-installed connectors 196; improperly terminated splitters 194), together with faults associated with the subscriber service drop 163C (e.g., pinched, kinked, and/or inferior quality/poorly shielded cable 163C; loose, corroded, or improperly-installed drop connectors 197A to the tap 188; improper/poor ground block splices 198), are conventionally deemed to account for 95% or more of ingress in the cable communication system (i.e., 75% inside subscriber premises plus 20% subscriber service drop, as noted above). While the hardline cable plant 180 generally is considered to be significantly better shielded and maintained (e.g., by the cable communication system owner/operator), in contrast the respective subscriber premises 190 typically are the least accessible and least controllable (i.e., they are generally private residences or businesses) and, as such, the least regularly-maintained portion of the cable communication system 160 (i.e., there is no regular access by the system owner/operator); hence, subscriber premises and their associated service drops are generally considered in the industry to be the most susceptible to signal leakage problems. Faults in subscriber service drops 163C and/or within subscriber premises 190 are considered to readily permit ingress from common ingress sources often found in household devices (e.g., appliances, personal computers, other consumer electronics, etc.) of cable communication system subscribers, as well as other ingress sources (e.g., garage door openers, various transmitting devices such as HAM or CB radios in the vicinity, fire and emergency communication devices, and terrestrial signals).
With respect to conventional ingress mitigation techniques, some approaches involve installing passive filters (e.g., in the taps 188 or within subscriber premises 190) to attenuate ingress originating from subscriber premises, while other approaches involve active systems that monitor communication traffic on the upstream path bandwidth and attenuate all or a portion of this bandwidth during periods of idle traffic. These approaches do not attempt to identify or eliminate ingress sources, but merely attempt to reduce their impact, and are accordingly not completely effective. Some other approaches, discussed in detail below, do attempt to identify subscriber-related faults that allow for ingress, but are generally labor and/or time intensive and largely ineffective. Furthermore, given the conventional presumption that 75% or more of ingress problems are deemed to relate to faults inside subscriber premises, even if ingress sources of this ilk are identified they may not be easily addressed, if at all (e.g., it may be difficult or impossible to gain access to one or more subscriber premises in which faults giving rise to ingress are suspected).
One conventional method for detecting ingress is to sequentially disconnect respective sections of hardline coaxial cable 163B (“feeders”) within the hardline cable plant 180 in which suspected ingress has been reported (e.g., by disconnecting a given feeder branch from the port of a directional coupler 189), and concurrently monitor resulting variations in the noise profile of the upstream path bandwidth as seen from the headend of the network (e.g., using the analyzer 110 shown in FIGS. 1, 2 and 4). This technique is sometimes referred to as a “divide and conquer” process (e.g., akin to an “Ariadne's thread” problem-solving process), and entails a significantly time consuming trial-and-error approach, as there are often multiple hardline coaxial cable feeder branches ultimately serving several subscriber premises, any one or more of which could allow for ingress to enter the network; accordingly, this technique has proven to be inaccurate and inefficient at effectively detecting points of ingress. Additionally, disruptive conventional methods involving disconnecting different feeder cables in the hardline cable plant cause undesirable subscriber interruption of ordinary services, including one or more of entertainment-related services, data and/or voice services, and potentially critical services (i.e. lifeline or 911 services).
Other conventional approaches to ingress mitigation employ low attenuation value switches (termed “wink” switches), installed in different feeder branches of the hardline cable plant, to selectively attenuate noise in the upstream path bandwidth and thereby facilitate localizing potential sources of ingress. Each wink switch has a unique address, and the various switches are sequentially controlled to introduce some amount of attenuation in the corresponding branch. The upstream path bandwidth is monitored at the headend (e.g., via the analyzer 110) while the wink switches are controlled, allowing observation at the headend for any changes in noise level in the upstream path bandwidth that may be attributed to respective corresponding branches. In one aspect, the use of wink switches in this approach constitutes an essentially automated methodology of the approach described immediately above (i.e., “divide and conquer”), but suffers from the same challenges; namely, the feeder branches being selectively attenuated ultimately serve several subscriber premises, any one or more of which could allow for ingress to enter the network. Accordingly, pinpointing potential points of ingress remains elusive.
In yet other conventional approaches, mobile transceivers may be employed in an attempt to detect both egress and ingress. For example, U.S. Pat. No. 5,777,662 (“Zimmerman”), assigned to Comsonics, Inc., discloses an ingress/egress management system for purportedly detecting both ingress and egress in a cable communication system. The system described in Zimmerman includes a mobile transceiver that receives RF egress and records GPS coordinates. The mobile transceiver also transmits a signal that is modulated with GPS coordinates. If there is a significant fault in the cable communication system allowing for ingress in the vicinity of signal transmission, the transmitted signal may be received at the headend of the network by a headend monitoring receiver. Based on transmitted signals that are received at the headend, a computer assigns coordinates to potential flaws within the cable system to generate a simple point map of same so that they may be repaired by a technician. One disadvantage of this system is that the transmitted signal modulated with GPS coordinates must be received at the headend with sufficient strength and quality to permit identification of the location of a potential flaw; in other words, if a potential flaw is not significant enough so as to admit the transmitted signal with sufficient strength, but is nonetheless significant enough to allow some amount of ingress to enter into the system, no information about the location of the potential flaw is received at the headend. Thus, obtaining an accurate and complete profile of potential ingress across a range of signal levels (and across a significant geographic area covered by a cable communication system), arguably is significantly difficult to achieve (if not impossible) using the techniques disclosed in Zimmerman.
It is generally understood that noise levels due to ingress in the upstream path bandwidth may vary as a function of one or more of time, frequency, and geographic location. Conventional ingress detection and mitigation techniques generally have been marginally effective in reducing ingress to some extent in the upper portion of the upstream path bandwidth (e.g., above 20 MHz); however, notable ingress noise levels continue to persist below approximately 20 MHz, with ingress noise at the lower end of this range (e.g., 5 MHz to approximately 18 MHz, and particularly below 16.4 MHz, and more particularly below 10 MHz) being especially significant.
As a result, it is widely accepted in the cable communication industry that only a portion of the upstream path bandwidth of a cable communication system, generally from about 20 MHz to 42 MHz, may be used in some circumstances (e.g., depending in part on the presence of broadband noise and/or narrowband interference, carrier frequency placement of one or more communication channels, carrier wave modulation type used for the channel(s), and channel bandwidth) for transmission of upstream information from subscriber modems to the headend, and that the lower portion of the upstream path bandwidth (e.g., generally from about 5 MHz to about 20 MHz, and particularly below 18 MHz, and more particularly below 16.4 MHz, and more particularly 10 MHz) is effectively unusable due to persistent ingress.
FIG. 12 shows an example of a power spectral density (PSD) (or “spectrum”) 2100A associated with the upstream path bandwidth 182 (i.e., 5 MHz to 42 MHz for the U.S.) of a conventional cable communication system, so as to illustrate the presence of ingress. The spectrum 2100A shown in FIG. 12 is provided as a screen shot from a display of a spectrum analyzer at the headend or coupled to the hardline cable plant 180 (e.g., serving as the analyzer 110 discussed above in connection with FIGS. 1, 2 and 4). In the spectrum analyzer screen shot of FIG. 12, the horizontal axis represents frequency in MHz, and the vertical axis represents signal level in dBmV.
As illustrated in FIG. 12, the presence of significant ingress disturbances 3500 from 5 MHz to just above approximately 20 MHz may be readily observed in the spectrum 2100A, including what appear to be a number of narrowband interference signals (also referred to as discrete “ingress carriers”) at approximately 6-7 MHz, 9 MHz, 10 MHz, 11.5 MHz, 13 MHz, 15 MHz, 18 MHz and 21 MHz, respectively (viewed from left to right in the screen shot). As noted above, constituent elements of such ingress disturbances 3500 possibly may be due to ham radio, short wave terrestrial signals, or other sources of narrowband interference that has entered via one or more faults. The spectrum 2100A in FIG. 12 also illustrates the presence of two channels 2103A and 2103B in the upstream path bandwidth 182, placed in a relatively “cleaner” portion of the spectrum 2100A at carrier frequencies of approximately 25 MHz and 30 MHz, respectively, wherein each upstream channel has a bandwidth 2109 of 3.2 MHz. From the relative signal levels of the channels 2103A and 2103B as compared to the ingress disturbances 3500, it may be readily appreciated from FIG. 12 that the ingress disturbances 3500 in the region of the spectrum 2100A below approximately 20 MHz essentially preclude the existence of any channels in this portion of the upstream path bandwidth.
FIG. 13 shows another example of a spectrum 2100B associated with the upstream path bandwidth 182 of a conventional cable communication system, so as to illustrate the presence of ingress in the form of broadband impulse noise (see Chapman, pages 90-91). As illustrated in FIG. 13, the presence of broadband impulse noise 3502 (indicated by a dashed oval in FIG. 13) covers a significant portion of the upstream path bandwidth 182 and is likely adversely impacting the transmission of upstream information via the channel 2103C (having a carrier frequency of approximately 27 MHz). As noted above, ingress sources giving rise to such broadband impulse noise 3502 include electric motors and power-switching devices (often found in household devices at subscriber premises).
FIG. 14 shows yet another example of a spectrum 2100C associated with the upstream path bandwidth 182 of a conventional cable communication system, so as to illustrate the presence of ingress. The screen shot of the spectrum 2100C shows a frequency marker at 50 MHz, around which frequency point (e.g., from about 40 MHz to 54 MHz) a significant “roll-off” may be readily observed in the spectrum 2100C (e.g., due to diplex filters included in amplifiers of the hardline cable plant), indicating the transition between the upstream path bandwidth 182 and the downstream path bandwidth (i.e., above 50 MHz). By conventional standards, the spectrum 2100C represents an example of a relatively “clean” upstream path bandwidth (see pages 31-33, FIG. 19 of “Digital Transmission: Carrier-to-Noise Ratio, Signal-to-Noise Ratio, and Modulation Error Ratio,” Ron Hranac and Bruce Currivan, Cisco whitepaper, November 2006, http://www.cisco.com/en/US/prod/collateral/video/ps8806/ps5684/ps2209/prod_white_paper0900aecd805738f5.html, hereafter “Hranac,” which whitepaper is hereby incorporated herein by reference in its entirety). In particular, the “noise floor” 2107 of the spectrum 2100C is relatively flat from about 25 MHz to 42 MHz. For purposes of the present disclosure, the noise floor of a spectrum refers to a measure of additive white Gaussian noise (AWGN) power (sometimes also referred to as “thermal noise” or “white noise”) within a measurement bandwidth of an instrument providing the spectrum. A noise floor may be substantially flat across a significant range of frequencies covered by the spectrum, or may vary within different frequency ranges of a given spectrum. In the example of FIG. 14, the noise floor 2107 is relatively flat from about 25 MHz to 40 MHz (there appears to be a very slight decrease in the noise floor over this range, but overall the noise floor is relatively flat in this range). Notwithstanding, below 25 MHz and particularly below 20 MHz, the noise floor rises significantly, and additionally the presence of significant ingress disturbances 3500 may be observed in the spectrum 2100C, including what appear to be a number of discrete “ingress carriers” at frequencies similar to those shown and discussed above in connection with FIG. 12.
While the particular regions of the spectrums 2100A and 2100C associated with ingress disturbances 3500 including discrete ingress carriers are particularly noteworthy in FIGS. 12 and 14, it should be appreciated that ingress may more generally impact the overall spectrum of the upstream path bandwidth; in particular, as illustrated in FIG. 13, various sources of ingress beyond the more discrete carriers shown amongst the ingress disturbances 3500 in FIGS. 12 and 14 may serve as wider-band noise sources that contribute to the overall noise profile of a spectrum, throughout significant portions (if not substantially all of) the spectrum.
The spectrum 2100C of FIG. 14 also illustrates the presence of an upstream channel 2103D in the upstream path bandwidth 182, wherein the upstream channel has a bandwidth 2109 of 3.2 MHz and is placed at a carrier frequency of 32.75 MHz (i.e., within a “cleaner” portion of the spectrum 2100B). A “carrier-to-noise ratio” (CNR) 2105 of the upstream channel 2103D also is indicated in FIG. 14; generally speaking, as discussed in greater detail below, a larger CNR for a channel (i.e., a greater distance between the average channel power, as represented by the top of the “haystack” profile for the channel, and the noise floor of the spectrum proximate to the channel) typically correlates with a reasonably functioning channel that effectively conveys upstream information, whereas relatively smaller values for CNR may be associated with channels that are not capable of conveying upstream information with sufficient reliability or accuracy. Accordingly, at least from a qualitative perspective, it may be appreciated from FIG. 14 that even in a so-called “clean” upstream spectrum by conventional standards, the presence of significant ingress disturbances 3500 in the region of the spectrum 2100C below approximately 20 MHz (notwithstanding the relatively lower magnitude of these disturbances as compared to FIG. 12) nonetheless poses significant challenges for the placement of appropriately functioning upstream channels in this region of the spectrum (e.g., see Table 4 and Table 5 discussed below).
Noise-Based Limitations on Cable System Communications
As noted above, noise that may be present on one or more physical communication media of a cable communication system may corrupt the integrity of information-bearing signals propagating on the media/medium. More specifically, as discussed above in connection with FIGS. 12 through 15, ingress in the upstream path bandwidth of a cable communication system can significantly (and adversely) impact the amount of usable spectrum within the upstream path bandwidth that can be employed to effectively convey upstream information from subscriber premises to the headend.
In the cable communication industry, various figures of merit are used to characterize the communication of information via modulated RF carrier waves (i.e., RF signals) in the presence of noise on the communication medium/media over which the RF signals propagate. A detailed treatment of such figures of merit is found in Hranac, referenced above.
One such figure of merit discussed in Hranac is referred to as “Carrier-to-Noise Ratio” (CNR), which is defined as the ratio of carrier or signal power to white-noise power in a specified bandwidth, as measured on a spectrum analyzer (or similar equipment). CNR often is expressed in units of decibels (dB), according to the relationship:
                                          CNR            ⁢                                                  ⁢                          (              dB              )                                =                      10            ⁢                                                  ⁢                          log              ⁡                              (                                                      P                                          carrier                      /                      signal                                                                            P                    noise                                                  )                                                    ,                            Eq        .                                  ⁢        7            where Pcarrier/signal is the carrier or signal power in Watts, and Pnoise is the additive white Gaussian noise (AWGN) power in Watts over a specified bandwidth. For digitally modulated RF signals (e.g., QPSK and QAM signals), the signal power Psignal is the average power level of the signal (also sometimes called average “digital channel power”) and is measured in the full occupied bandwidth of the signal (i.e., the symbol rate bandwidth, as discussed above in connection with Eq. 6).
With reference again to the upstream channel 2103D shown in FIG. 14, the spectrum of a typical QPSK or QAM channel resembles a “haystack” with an essentially flat top across the channel bandwidth. In the spectrum 2100C of FIG. 14, the height of the channel 2103 gives the signal density in units of dBmV as measured in the spectrum analyzer resolution bandwidth (RBW) (which in the example of FIG. 14 is 300 kHz). Given the specified channel bandwidth 2109 of 3.2 MHz, this RBW value can be scaled to the symbol rate bandwidth (i.e., 0.8×3.2 MHz=2.56 MHz) to arrive at the signal density in the channel (i.e., across the symbol rate bandwidth). Similarly, the height of the noise floor 2107 gives the noise density in units of dBmV as measured in the spectrum analyzer resolution bandwidth (RBW), and this also can be scaled to the symbol rate bandwidth to provide the total noise density in the channel. Given decibel units for the signal power and the noise power as expressed by the spectrum analyzer, the CNR is calculated by subtracting the total noise density in the channel from the signal density in the channel—however, since these respective values are both scaled similarly by the symbol rate bandwidth, this difference can be read directly from the spectrum analyzer screen shot as the vertical height between the average value at the top of the “haystack” channel spectrum to the noise floor, as shown by the reference numeral 2105 in FIG. 14 (In FIG. 14, the CNR 2105 is approximately 36 dB). This type of measurement of CNR from a spectrum analyzer screen shot is sufficiently accurate for CNR values greater than about 15 dB; if, however, the height between the top of the channel spectrum to the noise floor is between about 10 dB and 15 dB, an offset of about 0.5 dB should be subtracted from the observed height to provide a more accurate CNR measurement (for even smaller heights, larger offsets are required, e.g., subtracting as much as 1.5 dB from heights of about 5 dB).
Regarding channel power for upstream channels, and with reference again for the moment to FIG. 1, as discussed above a number of subscriber modems 165 that share a same physical communication channel in the upstream path bandwidth are referred to a “service group,” and the respective modulator circuits of these modems in the service group transmit upstream RF signals at different times (according to TDMA/ATDMA access protocols dictated by the CMTS 170 and the modems 165). Although transmitting at different times, subscriber modem upstream transmit levels are managed by the CMTS 170 so as to provide generally the same receive level for all subscriber modems (typically with less than 1 dB signal level difference among the modems) at a given demodulation tuner 174 of the CMTS tuned to demodulate the channel. Ingress (as well as AWGN) travels back to this same demodulation tuner, so the noise amplitude at the CMTS port coupled to the demodulation tuner is the same for all modems of the service group. Accordingly, the CNR for each modem in a service group (as observed at the CMTS port corresponding to the service group) typically is substantially similar if not virtually identical to other modems in the service group (unless there is a problem with a particular subscriber modem and/or an associated subscriber-related fault in a subscriber service drop or subscriber premises equipment within the service group).
Another related figure of merit discussed in Hranac is “Carrier-to-Noise-Plus-Interference Ratio” (CNIR), which makes a distinction between an essentially flat noise floor and more narrowband noise that could be present within the bandwidth of a physical communication channel. Rather than taking the ratio of the average channel power to only the white noise power in the symbol rate bandwidth, for CNIR the power of any narrowband interference present in the symbol rate bandwidth is added to the white noise power in the symbol rate bandwidth, and then the ratio of the channel power to this “noise-plus-interference power” is taken. The noise-plus-interference power may be measured during periods in which there is no RF signal being transmitted in the channel (“quiet times”); of course, it may be appreciated that for intermittent, random and/or bursty ingress sources, the noise-plus-interference power measurements may differ significantly as a function of time. In any event, the presence of significant interference power within the symbol rate bandwidth of a channel, in addition to white noise power, results in a CNIR significantly lower than a comparable CNR in the absence of such interference.
To illustrate the concept of CNIR relative to CNR, FIG. 15 shows the spectrum 2100C of FIG. 14 in which, for purposes of illustration, two additional hypothetical channels similar to the 3.2 MHz-wide channel 2103D are shown in “phantom” (with dashed lines to outline the channel) on the spectrum 2100C; in particular, a first hypothetical channel 2111 is placed at a center frequency of approximately 14 MHz, and a second hypothetical channel 2113 is placed at a center frequency of approximately 11 MHz. These two hypothetical channels occupy a portion of the spectrum 2100C in which the ingress disturbances 3500 are present, which disturbances would constitute a significant source of noise power in each of the hypothetical channels. In particular, it may be readily observed in FIG. 15 that the noise floor in the region of the spectrum corresponding to the two hypothetical channels 2111 and 2113 is notably higher than the noise floor 2107 in the vicinity of the channel 2103D (e.g., approximately 3 to 5 dB higher), and the respective peaks of the ingress carriers within the channels (constituting part of the ingress disturbances 3500) range from approximately 15 to 20 dB higher than the noise floor 2107; accordingly, the CNIR for each of the two hypothetical channels 2111 and 2113 would be significantly less than the CNR 2105 of the channel 2103D (e.g., CNIR on the order of 20-23 dB, as opposed to a CNR of 36 dB).
With reference again to the spectrum 2100A shown in FIG. 12, or the spectrum 2100B shown in FIG. 13, the situation for placing hypothetical channels in the region of the spectrum 2100A occupied by the ingress disturbances 3500, or the spectrum 2100B occupied by broadband impulse noise 3502, would be dramatically worse in terms of CNIR as compared to the situation discussed immediately above in connection with the relatively “cleaner” spectrum 2100C of FIG. 14. In FIG. 12, the respective CNRs of the channels 2103A and 2103B appear to be somewhat larger than the CNR 2105 for the channel 2103D shown in FIG. 14 (although the spectrums for the channels 2103A and 2103B in FIG. 12 may be obscuring some underlying narrowband interference noise, which indeed seems to be present to some extent in the region between the channels 2103A and 2103B, and just to the left of the channel 2103A). However, if the hypothetical channels of FIG. 15 where to be placed in the region of the spectrum 2100A in FIG. 12 that is occupied by the ingress disturbances 3500, or placed virtually anywhere within the spectrum 2100B of FIG. 13 below 27 MHz, the CNIR for these hypothetical channels would be severely lower than the CNR for the channels 2103A and 2103B (at one point around 10 MHz in the spectrum 2100A, there is an ingress carrier that is only about 3 dB below the channel power of the channels 2103A and 2103B).
On this qualitative basis alone, it would be readily appreciated from FIGS. 12 through 15 that the presence of significant broadband noise and narrowband interference signals in the region of the spectrum occupied by the ingress disturbances 3500 or the broadband impulse noise 3502, even for a relatively “clean” upstream spectrum as shown in FIG. 14, effectively precludes the placement of appropriately functioning channels in this region of the spectrum. As noted in Hranac, on page 6, DOCSIS specifies a minimum CNR for upstream channels, i.e., digitally modulated carriers, of 25 dB; this level of CNR, however, does not appear to be available for channels that would be placed below 20 MHz in the spectrums shown in FIGS. 12 through 15.
Table 4 below provides conventionally-accepted minimum carrier-to-in-channel noise values (CNR or CNIR, denoted generally as C/N) for a given physical communication channel that are required to support effective transport and demodulation of information carried over the channel using a particular modulation order of QAM; stated differently, different modulation orders of QAM require different minimum C/N values (e.g., see Chapman, page 38, Table 4 and page 133, Table 36; also see page 7, Table 3 of “The Grown-up Potential of a Teenage PHY,” Dr. Robert Howald et al., The NCTA 2012 Spring Technical Forum Proceedings, May 21, 2012, hereafter “Howald,” which publication is hereby incorporated herein by reference in its entirety; also see FIG. 1, page 150 of “256-QAM For Upstream HFC,” Thompson et al., NCTA 2010 Spring Technical Forum Proceedings, Los Angeles, Calif., May 2010, hereafter “Thompson,” which publication is hereby incorporated herein by reference in its entirety):
TABLE 4Uncoded TheoreticalOperator DesiredQAM Modulation OrderC/N (dB)C/N Target (db)QPSK (4-QAM)162216-QAM222832-QAM253164-QAM2834128-QAM 3137256-QAM 3440In Table 4, the “uncoded theoretical” values for C/N presume a bit error rate (BER) for demodulated symbols on the order of 10−8 (BERis the ratio of corrupt bits to total bits of information recovered from demodulation over some sampling period). Also, the uncoded theoretical values for C/N in Table 4 presume that no forward error correction or “FEC” (discussed in greater detail below) is employed in the transmission of information via a given physical communication channel (some MSOs tolerate a pre-FEC BER on the order of 10−7, although as noted above a pre-FEC BER on the order of 10−8 is more commonly adopted as a minimum BER threshold; modems typically start to have difficulty when BER is as high as on the order of 10−6 and modems typically fail to lock consistently when BER is as high as on the order of 10−5; for post-FEC BER, 10−9 is more commonly adopted as a minimum acceptable BER threshold for MSOs providing voice and/or data services, and more specifically “triple play” premium services).
The “operator desired” C/N targets listed in Table 4 are chosen to provide 6 dB of headroom above the uncoded theoretical values (to account for a wide variety of possible noise profiles that may occur in actual implementations with upstream information traffic from subscriber premises). From the C/N values provided in Table 4, it may be further appreciated in connection with FIGS. 12 through 15 that the presence of significant broadband noise and narrowband interference signals in the region of the spectrum occupied by the ingress disturbances 3500 or the broadband impulse noise 3502, even for a relatively “clean” upstream spectrum as shown in FIG. 14, effectively precludes the placement of appropriately functioning channels in this region of the spectrum having a QAM modulation order greater than 4 (i.e., only QPSK channels might function, if at all, in the lower portion of the upstream path bandwidth).
While CNR and CNIR (collectively C/N) provide illustrative figures of merit relating to RF signals as received by a demodulation tuner in the presence of noise (in terms of respective signal and noise powers), other instructive figures of merit relate to how effectively received signals are demodulated by a demodulation tuner of the CMTS. For example, other figures of merit quantify how effectively received RF signals are demodulated so as to recover the digital information transported by such signals, based on the symbol constellations associated with the modulation type and order used to generate the RF signals.
More specifically, as discussed above in connection with FIGS. 5 through 9 relating to QAM constellations and a QAM RF signal carrying upstream information (in which an amplitude A and phase φ of the signal is mapped to a particular point on the constellation corresponding to a symbol representing the upstream information), the presence of noise on a medium carrying such a signal may alter one or both of the signal's amplitude A and phase φ. Such alteration to the signal's amplitude A and/or phase φ may distort the signal such that, upon demodulation, the demodulated signal may be mapped not to the “target” constellation point corresponding to the original symbol carried by the signal, but instead to another neighboring point in the constellation, resulting in the wrong symbol being recovered by the demodulator.
FIG. 16 illustrates the effect of noise on the demodulation of QAM signals (e.g., as may be observed at a demodulation tuner 174 of the CMTS 170 at the headend 162—see FIG. 4) using an example of a QPSK or 4-QAM constellation diagram 5000A. As used in connection with FIG. 16, the term “decision boundary” refers to a bounded area on the constellation diagram 5000A, defined by certain ranges of values for I and Q, that are used to evaluate the constellation point/symbol of the constellation diagram to which a received and demodulated RF signal most appropriately maps. To this end, vertical and horizontal lines representing particular values of I and Q, respectively, are added to the constellation diagram 5000A so as to separate respective equidistant constellation points, thereby creating four squares in the constellation diagram, wherein a corresponding constellation point is at the center of each square. Each of the four squares constitutes a decision boundary for the corresponding constellation point/symbol, one of which decision boundaries is labeled in FIG. 16 with the reference numeral 5102 (i.e., the decision boundary in the top right hand corner of the constellation diagram 5000A).
Ideally, all received RF signals corresponding to a particular constellation point/symbol would, upon demodulation, map to the center of a decision boundary for the constellation point/symbol on the constellation diagram. However, various imperfections in the communication system, giving rise to the presence of noise within the signal's bandwidth and/or adversely affecting the propagation of the RF signal along the physical communication media in the system (i.e., the “channel response”), may cause the mapping of the signal upon demodulation to deviate from the center of the decision boundary. Based on the constellation's decision boundaries, a received RF signal having an amplitude A and a phase φ which upon demodulation provides I and Q values that fall within a particular decision boundary is deemed to represent the symbol corresponding to the constellation point within that decision boundary. Accordingly, such decision boundaries allow for a certain amount of fluctuation in the amplitude A and the phase φ of a received RF signal due to noise, without resulting in a demodulation error (or “symbol decoding error”).
In FIG. 16, this behavior may be observed by noting that within each decision boundary of the QPSK constellation diagram 5000A, there are multiple mapped points that fall outside of a shaded circle forming a small area around the center of each decision boundary. As more points are mapped to the constellation diagram over time, the mapped points form a “cloud” or “cluster” around the center of each decision boundary. Generally speaking, the amount of noise present within the channel and/or the channel response affects the overall spread of mapped points (spread of the “cloud” or “cluster”) within each decision boundary; in some cases, significant fluctuations in the amplitude A and/or the phase φ of a received RF signal due to noise in the channel and/or other adverse affect of the channel response may, upon demodulation, cause one or more points to be mapped close to or across a boundary edge, the latter resulting in a symbol decoding error. From FIG. 16, and with reference again to the various constellation diagrams shown in FIGS. 7 through 9 for different modulation orders of QAM, it may be appreciated that more dense constellations have smaller areas for the decision boundary associated with each constellation point/symbol; hence, as discussed above, RF signals based on higher order QAM modulation generally are more susceptible to noise-induced demodulation or symbol decoding errors.
To quantify the spread of mapped points in the example of FIG. 16, the first decision boundary 5102 shows a “target symbol vector” 5104, a “received symbol vector” 5106, and an “error vector” 5108 for one of several points mapped to this region of the constellation diagram 5000A. The target symbol vector 5104 represents an “ideal RF signal” that, upon demodulation, would be mapped to the center of the decision boundary 5102. The received symbol vector 5106 represents an actual signal that is received, demodulated, and mapped to the constellation diagram, and the error vector 5108 represents the difference between the received symbol vector 5106 and the closest target symbol vector 5104. As may be appreciated from FIG. 16, for each received demodulated signal mapped to the constellation diagram, a corresponding error vector may be determined based on the difference between the received symbol vector corresponding to the mapped point and the closest target symbol vector for the constellation point to which the signal is mapped. Thus, it may also be appreciated from FIG. 16 that smaller error vectors correspond to a more accurate mapping of received signals to the constellation diagram.
The term “modulation error ratio” (MER) refers to a figure of merit, based at least in part on error vectors similar to the error vector 5108 shown in FIG. 16, for quantifying the effectiveness of demodulating a received QAM RF signal. The MER takes into account any noise present within the channel bandwidth as well as the channel response, and provides a numerical metric to describe the spread (or “fuzziness”) of the “cloud” or “cluster” of points mapped to the constellation diagram around respective constellation points/symbols of the diagram.
Mathematically, MER is defined in terms of the average symbol power of some number N of symbols decoded from samples of demodulated received RF signals, divided by the average error power associated with the decoded symbols, according to the relationship:
                              MER          ⁢                                          ⁢                      (            dB            )                          =                  10          ⁢                                          ⁢                                    log              [                                                                    ∑                                          j                      =                      1                                        N                                    ⁢                                                                          ⁢                                      (                                                                  I                        j                        2                                            +                                              Q                        j                        2                                                              )                                                                                        ∑                                          j                      =                      1                                        N                                    ⁢                                                                          ⁢                                      (                                                                  δ                        ⁢                                                                                                  ⁢                                                  I                          j                          2                                                                    +                                              δ                        ⁢                                                                                                  ⁢                                                  Q                          j                          2                                                                                      )                                                              ]                        .                                              Eq        .                                  ⁢        8            Again, in Eq. 8 N denotes the total number of symbols determined from demodulation of successive samples of a received RF signal over a given time period, Ij and Qj respectively represent the in-phase and quadrature parts of the target symbol vector in the decision boundary to which the jth sample of the RF signal maps, and δIj and δQj respectively represent the in-phase and quadrature parts of the error vector corresponding to the actual point to which the jth sample of the RF signal maps. In practical application of Eq. 8, it is presumed that the measurement of MER is taken over a sufficiently large number N of samples (e.g., N>100) such that all of the constellation symbols are equally likely to occur; if this is the case, the numerator of Eq. 8 divided by N essentially represents the average symbol power of the constellation as a whole (which is a known constant for a given QAM modulation order). With this in mind, MER alternatively may be more generally defined as the ratio of average constellation symbol power to average constellation error power. In general, a higher value for MER represents a smaller cloud or tighter cluster of points for each symbol (less “fuzziness”), and corresponds to a lower level of impairments to the channel (e.g., noise and/or channel response anomalies) that may adversely impact propagation and hence demodulation of RF signals.
Another figure of merit that is closely related to MER is referred to as “error vector magnitude” (EVM). By convention, EVM is based on the root-mean-square (RMS) values of error vectors similar to the error vector 5108 shown in FIG. 16, as a percentage of the maximum symbol magnitude (e.g., corresponding to the constellation corner states). Accordingly, in contrast to MER, a lower EVM percentage value represents a smaller cloud or tighter cluster of points for each symbol (less “fuzziness”), and corresponds to a lower degree of impairment to the channel that may adversely impact demodulation of received RF signals. EVM is a linear figure of merit whereas MER is a logarithmic figure of merit, but both figures convey similar information regarding the effectiveness and accuracy of demodulation of QAM RF signals. For purposes of the discussion herein, MER is used primarily as the figure of merit regarding demodulation of QAM RF signals; however, it should be readily appreciated that MER may be converted to EVM, and vice versa, via well-known mathematical relationships (e.g., see Hranac, pages 28-29).
With reference again to FIGS. 1 and 4, whereas CNR or CNIR measurements for a given upstream channel often are made with the assistance of the analyzer 110 (e.g., a spectrum analyzer) at the headend 162 (or coupled to the hardline cable plant 180 as shown in FIG. 2), MER measurements typically are provided by the demodulation tuners 174 of a conventional CMTS 170 (or in some instances a specialized QAM analysis device/tool may be employed, e.g., the PathTrack™ HCU200 Integrated Return Path Monitoring Module offered by JDSU, see http://www.jdsu.com/ProductLiterature/hcu200_ds_cab_tm_ae.pdf). A detailed discussion of demodulation tuner functionality and the manner in which MER measurements may be made is discussed in Hranac (e.g., see Hranac, pages 19-21).
Recall from the discussion above in connection with FIGS. 1 and 4 that multiple subscriber premises typically share a single upstream channel as a “service group.” In exemplary implementations of conventional demodulation tuners 174 and subscriber modems 165 according to the DOCSIS standard utilizing TDMA or ATDMA, different subscriber premises of a given service group transmit their corresponding portions of upstream information to the headend as transmission “bursts” with some preordained timing (in this context, a demodulation tuner 174 also is sometimes referred to as a “burst receiver”). Typical demodulation tuners of a CMTS are configured to report various operating parameters, including MER (some instruments refer to MER as “signal-to-noise ratio” (SNR) as well as “receive modulation error ratio” (RxMER)). Some CMTSs are configured to report MER on both a per-channel basis and a per-subscriber-modem basis, in which per-channel MER measurements provide an average MER value over some number of valid bursts from the service group (DOCSIS specifies upstream MER measurements as an estimate provided by a CMTS demodulation tuner of the ratio of average constellation energy with equally likely symbols to average squared magnitude of error vectors, over some number of valid bursts from different subscriber modems of the service group sharing the physical communication channel assigned to the demodulation tuner). Accordingly, unless otherwise indicated herein, any reference to numerical MER values represents an MER value for a physical communication channel, rather than for a particular subscriber modem (that may be sharing the channel with other modems of the service group).
Various other features that may be implemented in demodulation tuners 174 of the CMTS 170 and subscriber modems 165 adopting DOCSIS version 1.1 and higher also may need to be considered in connection with MER measurements provided by a given demodulation tuner (or QAM analysis device), as they may provide “processing gains” to improve channel performance in the presence of noise or other channel impediments (e.g., see “Advanced Physical Layer Technologies for High-Speed Data Over Cable,” Cisco Whitepaper, August 2005, http://www.cisco.com/en/US/prod/collateral/video/ps8806/ps5684/ps2209/prod_white_paper0900aecd8066c6cc_ps4969_Products_White_Paper.html, hereafter “Cisco Advanced PHY,” which whitepaper is incorporated by reference herein in its entirety; also see “QAM Overview and Troubleshooting Basics for Recently digital Cable Operators,” JDSU whitepaper, October 2009, http://www.jdsu.com/ProductLiterature/Digital_QAM_Signals_Overview_and_Basics_of_Testing.pdf, hereafter “JDSU QAM Overview,” which whitepaper is hereby incorporated by reference herein in its entirety; also see Hranac, pages 18-21). For example, subscriber modems 165 may implement an adaptive equalizer (sometimes referred to as a “pre-equalizer”) to intentionally distort the waveform of a transmitted upstream RF signal so as to compensate for the upstream channel frequency response; similarly, demodulation tuners may implement a complimentary adaptive equalizer to compensate for channel response effects (e.g., group delay variation, amplitude slope or ripple, and/or microreflections). While the adaptive equalizers in one or both of the subscriber modem and the demodulation tuner may significantly improve the channel response for a given channel over which upstream information is being communicated, there are practical limitations on the extent to which channel impairments can be compensated. Notwithstanding, some demodulation tuners of conventional CMTSs (as well as various QAM analysis devices) provide “equalized” MER measurements (i.e., for which a given demodulation tuner adapts its equalizer on each data traffic burst received from a subscriber modem of the service group), as well as “unequalized” MER measurements (e.g., for which adaptive equalization is bypassed or not enabled). For purposes of the present disclosure, unless otherwise indicated specifically, any reference to numerical MER values herein presumes an “unequalized” MER measurement (which is typically a lower numerical value than a corresponding equalized MER measurement for the same channel, absent the processing gain provided by adaptive equalization).
Additionally, some conventional demodulation tuners of CMTSs are equipped with “ingress cancellation” circuitry, which is designed to attenuate to some degree in-channel narrowband interference arising from ingress (as discussed above in the previous section). Such circuitry is similar to the adaptive equalizers discussed immediately above, in that it dynamically detects and measures in-channel narrowband interference and adapts filter coefficients so as to try to attenuate the detected/measured interference. Ingress cancellation circuitry may add some broadband white noise (AWGN) to the channel; additionally, ingress cancellation circuitry generally is only effective at mitigating some degree of demodulation error due to narrowband interference in a channel that is already “minimally functioning” (e.g., channels having a CNIR above a certain minimum threshold, in which there may be a single narrowband ingress carrier having relatively modest peak power within the channel bandwidth, which may be further attenuated by the ingress cancellation circuitry).
As noted above, however, the presence of appreciable interference (e.g., in the form of multiple ingress carriers of significant strength—see FIGS. 12 and 14; and/or broadband impulse noise—see FIG. 13) in some portion(s) of the spectrum of the upstream path bandwidth (e.g., below 20 MHz, and particularly below 18 MHz, and more particularly below 16.4 MHz, and more particularly below 10 MHz) may significantly impair, or preclude the existence of, a channel within that/those portion(s) of the spectrum, even with demodulation tuners that employ ingress cancellation circuitry. Indeed it has been noted that while ingress cancellation circuitry is effective at facilitating reliable transmissions in the middle to high end of the upstream path bandwidth, in contrast ingress cancellation circuitry generally is not effective below 20 MHz, where channels are most vulnerable to broadband impulse noise (and multiple significant ingress carriers) (e.g., see Chapman, page 69; also see Cisco Advanced PHY, pages 23-28, in which tests of ingress cancellation circuitry for a 16-QAM 3.2 MHz bandwidth channel below 20 MHz result in significant bit error rates (BER)—ping losses of 0.01% and higher, suggest BERs on the order of 10−4 to 10−6, i.e., notably worse BERs than conventionally acceptable post-FEC BER on the order of 10−9; also see Thompson, pages 148-149—“Laboratory Measurements”).
With respect to the effect of ingress cancellation circuitry on MER, assuming a minimally functioning channel (e.g., pre-FEC BER on the order of no higher than 10−7 and preferably on the order of 10−8), MER measurements for a given channel in which ingress cancellation circuitry is employed generally are higher than CNIR measurements for the channel, due to some degree of attenuation of limited narrowband interference. In any event, for purposes of the present disclosure, the function of ingress cancellation circuitry, if present in a demodulation tuner of the CMTS, is treated similarly to adaptive equalization; accordingly, unless otherwise indicated specifically, again any reference to numerical MER values herein presumes an “unequalized” MER measurement (in which neither ingress cancellation circuitry nor other adaptive equalization is employed in connection with demodulation of a received upstream RF signal). Again, as noted above, an unequalized MER measurement for a channel typically provides a lower numerical value than a corresponding equalized MER measurement in which ingress cancellation and/or adaptive equalization is employed.
With the foregoing in mind, there are conventionally-accepted approximate minimum unequalized MER values (“MER failure threshold values”) for a given physical communication channel that are required to support effective transport and demodulation of information carried over the channel using a particular modulation order of QAM; stated differently, different modulation orders of QAM require different MER failure threshold values (e.g., see “Broadband: Equalized or Unequalized? That is the Question,” Ron Hranac, Communication Technology, Feb. 1, 2007, http://www.cable360.net/print/ct/operations/bestpractices/21885.html, hereafter “Hranac 2007,” which article is hereby incorporated by reference herein in its entirety; also see Hranac, page 23, Table 4; also see JDSU QAM Overview, pages 7-8). Table 5 below provides representative unequalized MER failure threshold values to support different modulation orders of QAM, and corresponding C/N target values (from Table 4 above).
TABLE 5QAM modulation orderMinimum Unequalized(symbols per constellation)MER (dB)C/N Target (db)QPSK (4-QAM)10-132216-QAM17-202864-QAM22-2434256-QAM 28-3040Hranac and Hranac 2007 suggest that unequalized MER values in an operational system should be kept at least 3 dB to 6 dB above the failure threshold for the modulation type in use (also see Hranac, page 23, footnote 11, which notes that many cable operators use the following unequalized MER values as minimum operational values: QPSK ˜18 dB; 16-QAM ˜24 dB; 64-QAM ˜27 dB; and 256 QAM ˜31 dB; also see Chapman, page 77, Table 18, which notes an equalized MER of 37 dB to support a pre-FEC error-free 256-QAM channel in the presence of AWGN, and without ingress carriers; also see “Pushing IP Closer to the Edge,” Rei Brockett et al., The NCTA 2012 Spring Technical Forum Proceedings, May 21, 2012, which publication is hereby incorporated herein by reference in its entirety—page 5 notes an MER of 37 dB sufficient to support a modulation order of 256-QAM).
According to Hranac, CNR, CNIR and MER for a given channel should be virtually identical in an “ideal” system with no impairment to the channel other than additive white Gaussian noise (AWGN) and with full traffic loading (e.g., see Hranac, page 39); in many practical implementations, however, Hranac suggests that for CNR values of between 15-25 dB, again where AWGN is the primary channel impairment (i.e., CNR=CNIR), MER should agree with CNR to within about 2 dB or less. Hranac also points out that the MER of a channel is less than, or at best equal to, CNR, but never greater than CNR, and that MER may be appreciably less than CNR if significant impairments to the channel beyond AWGN exist (e.g., ingress disturbances and/or channel response effects such as group delay variation, amplitude slope or ripple, and/or microreflections). Accordingly, it should be appreciated that the CNR for a given channel may appear to be relatively high while at the same time the MER for the channel can be unusually low—however, the converse can not be true; stated differently, a channel may have a relatively high CNR and low MER, but a channel with a low CNR (or low CNIR) will always have a correspondingly low MER.
With the foregoing also in mind, and with reference again to FIGS. 12 and 14, the presence of significant ingress disturbances 3500 in the spectrums 2100A and 2100C are indicative of notably low prospective CNIRs for hypothetical channels placed in this portion of the upstream path bandwidth, as discussed above in connection with FIG. 15. Accordingly, in view of the relationship between MER and CNR/CNIR discussed immediately above, and the MER fault thresholds given in Table 5 above, such low prospective CNIRs due to significant ingress disturbances essentially serve as a “non-starter” for implementation of functioning channels in this portion of the upstream path bandwidth.
As discussed briefly above, one technique specified in DOCSIS to improve the robustness of information transmission over a physical communication channel in the presence of broadband impulse or burst noise is referred to as “Forward Error Correction” (FEC). FEC typically involves the transmission of additional data (sometimes referred to as “parity bytes” or “overhead”), together with data packets constituting upstream information from subscriber premises that is encoded on a channel's modulated carrier, to allow for the correction of bit errors post demodulation. FEC conventionally is accomplished by adding redundancy to the transmitted information using a predetermined algorithm. Unlike adaptive equalization and ingress cancellation circuitry, FEC does not affect MER measurements; rather, if bit errors do occur as a result of demodulation, FEC provides a technique by which such bit errors may be corrected based on the “overhead” or redundancy built into the information transported over the channel. In this manner, using FEC as part of the modulation and demodulation processing of information transported over a physical communication channel in which a given modulation order of QAM is employed may permit somewhat less stringent C/N or MER threshold values for ensuring reliable channel operation (e.g., the C/N and MER threshold values shown in Table 4 and Table 5 may be somewhat lower if FEC is employed).
To illustrate the foregoing premise, Table 6 below provides additional C/N performance targets for different modulation orders of QAM using two different types of FEC, namely, “Reed-Solomon” FEC (commonly employed in conventional subscriber modems and CMTSs, and “Low Density Parity Check” (LDPC) codes, currently proposed for future “next generation” implementations of DOCSIS-compliant modulators and demodulators; see Chapman, pages 119 through 126). For comparison, Table 6 also includes uncoded theoretical C/N values from Table 4 above. From Table 6, it may be appreciated that employing these forms of FEC provides additional robustness against channel impairments, thereby permitting lower C/N performance targets for sustaining functional channels (e.g., see Chapman, FIG. 57, page 132, and Table 36, page 133).
TABLE 6UncodedReed-SolomonQAM ModulationTheoreticalFECLDPC CodedOrderC/N (dB)C/N Target (dB)C/N Target (dB)QPSK (4-QAM)16104 16-QAM221610 32-QAM251913 64-QAM282216 128-QAM312519 256-QAM342822 512-QAM3731251024-QAM4034282048-QAM4337314096-QAM464034
While providing some degree of enhanced protection against noise-induced errors, FEC does not work well however if significant impulse noise creates many demodulation errors in succession (e.g., see “Upstream FEC Errors and SNR as Ways to Ensure Data Quality and Throughput,” Cisco Whitepaper, Document ID: 49780, Oct. 4, 2005, which whitepaper is incorporated by reference herein in its entirety). In particular, ingress types that could introduce errors that are uncorrectable via FEC include excessive impulse noise and/or narrowband interference (e.g., ingress carriers). As discussed earlier in connection with the C/N metrics presented in Table 4, some MSOs tolerate a pre-FEC BER on the order of 10−7, although a pre-FEC BER on the order of 10−8 is more commonly adopted as a minimum BER threshold; for channels in which FEC is employed, a BER on the order of 10−9 is more commonly adopted as a minimum acceptable BER threshold. In particularly noisy environments, however, even with FEC, this level of BER may be challenging if not impossible to attain.
In addition to RS-FEC, DOCSIS versions 2.0 and higher support an alternative to the TDMA and ATDMA time-division protocols (employed by multiple subscriber modems to share a channel as a service group) that purportedly is more robust than its TDMA/ATDMA counterparts against channel impairments such as broadband impulse noise; as briefly noted above, this alternative protocol is commonly known as Synchronous Code Division Multiple Access (S-CDMA) (see Chapman, pages 88 through 95). In S-CDMA, each symbol of data is multiplied at the modulation tuner (i.e., transmitter) of a subscriber modem by a spreading code including some number of codes, which spreads out each symbol in the time domain by as much as 128 times. Accordingly a noise burst that may wipe out many QAM symbols being transported over an ATDMA channel must have a significantly longer duration to have the same effect on an S-CDMA channel. At the same time, there is no reduction in data throughput in an S-CDMA channel, as multiple subscriber modems of a service group may transmit at the same time (the orthogonal spreading code is used to differentiate respective transmissions from different modems, which may transmit simultaneously). While S-CDMA is widely touted as a possible solution for deployment of channels in the troublesome portion of the upstream path bandwidth below 20 MHz, it has nonetheless remained largely unused in practice by MSOs despite its availability in DOCSIS 2.0 and DOCSIS 3.0 certified equipment (e.g., see Chapman, page 89).
Regarding the figures of merit discussed above, the power of an S-CDMA burst depends on the number of codes used in the spreading code; accordingly, S-CDMA presents some challenges for accurate CNR and MER measurements at the headend. For purposes of the present disclosure, unless otherwise specifically stated, it is presumed that any numerical CNR, CNIR, or MER values provided herein are associated with channels implemented according to TDMA or ATDMA protocols, and not S-CDMA (i.e., it is presumed that subscriber modems and demodulation tuners of the CMTS are not employing S-CDMA unless specifically noted otherwise).
Given the accepted limitations on the upstream path bandwidth arising from ingress, conventional cable communication systems implement “channel plans” for the upstream path bandwidth that attempt to avoid the various challenges posed by the presence of ingress (as well as other potential channel impairments). For purposes of the present disclosure, a “channel plan” for the upstream path bandwidth of a cable communication system refers to the designation of one or more of: 1) a number of channels occupying the upstream path bandwidth; 2) the carrier frequency/frequencies at which the channel(s) is/are placed in the upstream path bandwidth; 3) the bandwidth(s) of the channel(s); 4) the QAM modulation order(s) of the channel(s); 5) an operational average power level for each channel with respect to the overall power budget of the upstream path; and 6) an aggregate deployed or “raw” data rate (deployed capacity) for the channel plan. For example, with reference again to FIG. 12, the channel plan for the upstream path bandwidth 182 represented by the spectrum 2100A includes the two channels 2103A and 2103B, wherein the channel 2103A has a carrier frequency of about 25 MHz and the channel 2103B has a carrier frequency of about 30 MHz, both channels have a bandwidth of 3.2 MHz, and both channels use QPSK (4-QAM) for modulation of upstream information conveyed by the channel. For a given upstream channel plan, an aggregate upstream raw data rate (i.e., “deployed capacity”) may be calculated based on Table 2 and the bar graph shown in FIG. 10. In particular, for the channel plan shown in FIG. 12, FIG. 10 indicates that a 3.2 MHz wide QPSK channel has a raw data rate of approximately 5 Mbits/s (and from Table 2, for a QPSK channel, 1.6 bps/Hz×3.2 MHz=5.12 Mbits/s); therefore, given two channels having a raw data rate of approximately 5 Mbits/s each, the deployed upstream capacity of the channel plan shown in FIG. 12 is approximately 10 Mbits/s.
Generally speaking, only a portion of the deployed capacity for a given physical communication channel, or for a given channel plan, is available for transporting upstream information from one or more subscriber premises. In particular, if forward error correction (FEC) is employed (which is effectively a given in conventional communication systems), some of the available deployed capacity is used for the overhead involved in FEC; similarly, the transmission of upstream information in bursts of data, and formulation of data into IP data packets, also involves some administrative overhead that consumes some of the deployed capacity of a given channel/channel plan.
More specifically, the “overhead” or “parity bytes” that are employed in FEC understandably take up a portion of the deployed or “raw” data rate of a physical communication channel that would otherwise be used for upstream information from one or more subscriber premises. For example, in conventional Reed-Solomon FEC implementations, k represents the number of data symbols being encoded in a given block of data, n represents the total number of coded symbols in the encoded block, and t represents the symbol-error correcting capability of the code, where n−k=2t provides the number of parity symbols constituting the “overhead” (i.e., to ensure correction of 8 erroneous symbols, 16 parity symbols of “overhead” is required). Thus, a “code rate” of the FEC, i.e., the portion of the encoded data block effectively constituted by the original k symbols of upstream information being encoded, is given by k/n. Example code rates for FEC commonly employed in conventional cable communication systems are on the order of approximately 0.7 to 0.9; for example, consider an RS-FEC in which k=100, n=116, and t=8, providing for an FEC code rate of 0.862 (e.g., see Chapman, pages 122-123). Accordingly, the “raw” data rate of a physical communication channel employing FEC is “discounted” or “de-rated” by an amount corresponding to the FEC code rate (in the foregoing example, approximately an 86% de-rating factor).
There are additional aspects of “DOCSIS overhead,” beyond FEC, that further limit an effective data rate of a physical communication channel. For example, DOCSIS “physical layer overhead” (“PHY overhead”) relates to some number of symbols in a transmission burst received at the CMTS that are dedicated to a preamble and a guard band, thereby diminishing the number of symbols in a burst relating to the actual information payload. In exemplary implementations, 40 symbols of a 2048 symbol burst may be used for PHY overhead, thereby further reducing an effective data rate for the channel by a factor of 0.9805. In addition, DOCSIS “media access and control layer overhead” (“MAC overhead”) relates to the number of bytes of information typically included in IP data packets, packet header sizes, number of headers in a given transmission burst, and other factors that further reduce an effective data rate for the channel, typically by a factor of approximately 0.91 (e.g., see Chapman, pages 122-123).
Accordingly, when one considers the cumulative effect of FEC, PHY overhead, and MAC overhead on the data rate of a physical communication channel, the deployed or “raw” data rate of the channel needs to be discounted or de-rated to provide an effective data rate for upstream information from one or more subscriber premises that is conveyed over the channel. Given the examples provided above for FEC, PHY Overhead, and MAC Overhead, a representative de-rating factor is on the order of (0.862)×(0.985)×(0.91)≈0.77, or approximately 77% of the deployed or raw data rate for the channel. In view of the foregoing, for purposes of the present disclosure, an “effective data rate” for a physical communication channel (or similarly an “effective upstream capacity” for a channel plan) takes into consideration a cumulative de-rating factor that is applied to the “raw” data rate of the channel (or the deployed capacity of the channel plan), wherein the de-rating factor may relate at least in part to FEC overhead and/or DOCSIS overhead. For purposes of the present disclosure, unless otherwise indicated in illustrative examples, a representative FEC/DOCSIS overhead de-rating factor of approximately 0.8 (80%) of the raw data rate of a channel (or the deployed capacity of a channel plan) is presumed to determine an effective data rate of the channel (or the effective upstream capacity of the channel plan).
FIG. 17 illustrates a chart showing a typical DOCSIS upstream channel plan 2000A for a conventional cable communication system (e.g., see page 6 of “Better Returns from the Return Path Implementing an Economical Migration Plan for Increasing Upstream Capacity,” Brian O'Neill and Rob Howald, Motorola whitepaper, September 2008, http://www.motorola.com/staticfiles/Video-Solutions/Solutions/Industry %20Solutions/Service %20Providers/Cable %20Operators/Broadband %20Access %20Networks %20(BAN)/Fiber %20Deep/_Documents/Static %20files/Better %20Returns %20from %20the %20Return %20PathWhitepaper92008.pdf, hereafter referred to as “Motorola,” which whitepaper is hereby incorporated by reference herein in its entirety). In the chart of FIG. 17, the horizontal axis represents frequency within the upstream path bandwidth from 5 MHz to 42 MHz, and the vertical axis indicates the QAM modulation order of a given channel and the associated data rate in Mbits/s/MHz (from Table 2 above). As shown in FIG. 17, Motorola indicates that a typical upstream channel plan 2000A includes two channels 2002A and 2002B having respective carrier frequencies generally in a range around 30 MHz to 35 MHz (e.g., well above 20 MHz, below which ingress becomes a salient issue, and well below the diplex filter roll-off of the upstream path bandwidth at 42 MHz). Each of the channels 2002A and 2002B has a bandwidth of 3.2 MHz, and uses a QAM modulation order of 16 (16-QAM). In one aspect, the selection of the QAM modulation order of 16 for the channels 2002A and 2002B, as well as the appropriate channel carrier frequencies, is based at least in part on prevailing noise and/or channel response conditions typically expected in a conventional cable system that give rise to a particular MER failure threshold value according to Table 5 above (i.e., in the case of 16-QAM, an MER of at least 17 to 20 dB, and preferably 24 dB). The deployed upstream capacity for the channel plan 2000A may be estimated from FIG. 17 by considering the two channels, each having a bandwidth of 3.2 MHz and respective data rates of 3.2 Mbits/s/Hz; i.e., 2×3.2×3.2≈20.5 Mbits/s. Using a de-rating factor of approximately 80%, this deployed upstream capacity corresponds to an effective upstream capacity of approximately 16.4 Mbits/s for the channel plan 2000A.
FIG. 18 illustrates a chart showing another DOCSIS upstream channel plan 2000B proposed by Hranac and including a 64-QAM upstream channel (e.g., see “Broadband: Another Look at Upstream 64-QAM,” Ron Hranac, Communications Technology, Apr. 1, 2009, http://www.cablefax.com/ct/operations/bestpractices/Broadband-Another-Look-at-Upstream-64-QAM—34894.html, hereafter referred to as “Hranac Broadband”). In particular, in the proposed channel plan 2000B, a first channel 2002C has a carrier frequency of 21.6 MHz, a bandwidth of 6.4 MHz, and uses 64-QAM, and a second channel 2002D has a carrier frequency between 30-35 MHz (e.g., approximately 32.5 MHz), a bandwidth of 3.2 MHz, and uses 16-QAM. For the proposed upstream channel plan 2000B, Hranac Broadband also suggests multiple narrower bandwidth QPSK channels (e.g., three 1.6 MHz-wide QPSK channels 2002E) placed between 35 MHz and approximately 40 MHz. From FIG. 18 and Table 2, and with the aid of Table 7 below, it can be seen that the proposed channel plan 2000B has a deployed upstream capacity of just under approximately 50 Mbits/s (corresponding to an effective upstream capacity of approximately 39 Mbits/s using an 80% de-rating factor):
TABLE 7Channel BWRaw Data RateChannel QAM(MHz)(Mbits/s)646.430.72163.210.244 (QPSK)1.62.564 (QPSK)1.62.564 (QPSK)1.62.56TOTAL DEPLOYED CAPACITY48.64
In proposing the channel plan 2000B of FIG. 18, Hranac Broadband notes that the wider bandwidth 64-QAM channel should be placed below 30 MHz to avoid potential channel impairments in the form of group delay arising from diplex filters associated with one or more amplifiers of the hardline cable plant. In particular, as upstream RF signals travel from subscriber modems toward the headend, these upstream signals often pass through multiple amplifiers of the hardline cable plant, each of which amplifiers contains diplex filters to only permit passage of RF signals within the upstream path bandwidth. Hranac Broadband notes that, although the roll-off of such filters in the area of 42 MHz (as discussed above in connection with the spectrum 2100C shown in FIG. 14) is generally considered to only affect signal transmission above 35 MHz, this roll-off nonetheless may adversely affect reliable transmission of signals modulated with more dense constellations (e.g., 64-QAM), due to diplex filter-related group delay (which is a cumulative effect that is exacerbated with passage of the signal through a greater number of amplifiers/filters). Accordingly, Hranac Broadband suggests placement of the 64-QAM channel 2002C below 30 MHz, and using lower modulation order and lower bandwidth channels above 30 MHz (e.g., the channels 2002D and 2002E shown in FIG. 18).
FIG. 19 illustrates a chart showing yet another DOCSIS upstream channel plan 2000C proposed by Motorola (see Motorola, page 6) that suggests three upstream 64-QAM channels above 20 MHz, and two 16-QAM channels below 20 MHz. The channel plan 2000C proposed by Motorola, however, requires the use of the S-CDMA protocol (as opposed to TDMA or ATDMA) for at least two and possibly three of the upstream channels. In particular, in the channel plan 2000C, two channels collectively labeled as 2002F are proposed below 20 MHz, each having a bandwidth of 3.2 MHz and using 16-QAM with S-CDMA (the requirement of S-CDMA is indicated in FIG. 19 using cross-hatching for the channels 2002F). Motorola acknowledges that, in the upstream path bandwidth, frequencies below about 20 MHz, and especially below 15 MHz, tend to be cluttered with interference and impulse noise (e.g., see the ingress disturbances 3500 and broadband impulse noise 3502 shown in FIG. 12-16 as discussed above), thus making this band unsuitable for DOCSIS performance and unused for DOCSIS services. Motorola nonetheless proposes that lower modulation order QAM (e.g., 16-QAM), coupled with S-CDMA technology, could be used for channels below 20 MHz.
The proposed channel plan 2000C shown in FIG. 19 also indicates the placement of a 6.4 MHz-wide 64-QAM channel 2002G above 20 MHz. Motorola indicates that this channel may be implemented successfully using an ATDMA protocol, but also indicates, however, that this channel may require the use of S-CDMA technology for successful implementation (accordingly, the possible requirement of S-CDMA for the channel 2002G is indicated with “confetti-type” fill in FIG. 19). Motorola further suggests that the proposed channel plan 2000C also may include two 6.4 MHz-wide 64-QAM channels 2002H employing ATDMA (i.e., without requiring the use of S-CDMA). It is noteworthy that the placement of these channels 2002H as recommended by Motorola would occupy a portion of the upstream path bandwidth well above 30 MHz, in stark contrast to the teachings of Hranac Broadband (which advise against placement of 64-QAM channels above 30 MHz, at least in part due to undesirable diplex filter-related group delay effects). In any event, from FIG. 19 and Table 2, and with the aid of Table 8 below, it can be seen that the channel plan 2000C, which would require the use of S-CDMA for at least two if not three of the specified channels, has a deployed upstream capacity of just over approximately 110 Mbits/s (corresponding to an effective upstream capacity of approximately 90 Mbits/s using an 80% de-rating factor):
TABLE 8Channel BWRaw Data RateChannel QAM(MHz)(Mbits/s)16 (S-CDMA)3.210.2416 (S-CDMA)3.210.24 64 (S-CDMA?)6.430.72646.430.72646.430.72TOTAL DEPLOYED CAPACITY112.64
FIG. 20 illustrates a chart showing yet another DOCSIS upstream channel plan 2000D proposed by Chapman (see Chapman, page 89, FIG. 37) that suggests a total of seven channels, i.e., four ATDMA channels above 22.8 MHz and three S-CDMA channels below 22.8 MHz; accordingly, like the channel plan 2000C shown in FIG. 19 proposed by Motorola, Chapman's proposed channel plan also requires S-CDMA channels below 20 MHz. Specifically, below 22.8 MHz, Chapman's channel plan 2000D calls for a 32-QAM S-CDMA channel 20021 having a bandwidth of 3.2 MHz, another 32-QAM S-CDMA channel 2002J having a bandwidth of 6.4 MHz, and a 64-QAM S-CDMA channel 2002K having a bandwidth of 6.4 MHz. Above 22.8 MHz, Chapman's proposed channel plan 2000D calls for two 64-QAM ATDMA channels 2002L each having a bandwidth of 6.4 MHz, a 64-QAM ATDMA channel 2002M having a bandwidth of 3.2 MHz, and a 16-QAM ATDMA channel 2002N having a bandwidth of 3.2 MHz (notably, Chapman's proposed channel plan 2000D adopts in part the recommendation of Hranac Broadband against placement of 64-QAM channels above 30 MHz, by using the smaller bandwidth 16-QAM channel 2002N closest to the diplex filter roll-off at the highest end of the spectrum).
From FIG. 20 and Table 2, and with the aid of Table 9 below, it can be seen that the channel plan 2000D, which would require the use of S-CDMA for all channels below 22.8 MHz, has a deployed upstream capacity of just over approximately 155 Mbits/s (corresponding to an effective upstream capacity of approximately 125 Mbits/s using an 80% de-rating factor):
TABLE 9Channel BWRaw Data RateChannel QAM(MHz)(Mbits/s)32 (S-CDMA)3.212.832 (S-CDMA)6.425.664 (S-CDMA)6.430.72646.430.72646.430.72643.215.36163.210.24TOTAL DEPLOYED CAPACITY156.16
Although the proposed channel plan 2000C of FIG. 19 and proposed channel plan 2000D of FIG. 20, both employing multiple S-CDMA channels, promise deployed upstream capacities on the order of 110 Mbits/sec and 155 Mbits/sec, respectively, industry commentators have noted that such plans have not been effectively adopted by operators of conventional cable communication systems to achieve this degree of upstream capacity. In particular, while S-CDMA has been available since the advent of DOCSIS version 2.0 in 2002, there has not been significant adoption of S-CDMA technology by MSOs, in part due to some required improvements in software and algorithms employed by demodulation tuners of the CMTS to decode S-CDMA, and appropriate selection of spreading codes given the variety of possible noise profiles that may be encountered in the upstream path bandwidth (e.g., see “Moto: S-CDMA Starting to Spread,” Jeff Baumgartner, Light Reading Cable, Feb. 16, 2010, http://www.lightreading.com/document.asp?doc_id=187997&site=lr_cable; also see Chapman, page 89).
Industry commentators also have noted that, even with S-CDMA and the additional upstream channel bonding capabilities available with the advent of DOCSIS version 3.0, at present the maximum effective upstream capacity arising from current implementations by some MSOs of upstream channel plans under DOCSIS version 3.0 is on the order just under 100 Mbits/sec at best (e.g., see “Cisco Hints at What Comes After Docsis 3.0,” Jeff Baumgartner, Light Reading Cable, May 14, 2012, http://www.lightreading.com/document.asp?doc_id=220896&site=lr_cable&). This arises from the channel plan 2000E shown in FIG. 21, which includes four 64-QAM ATDMA channels 2002P each having a bandwidth of 6.4 MHz and having carrier frequencies of approximately 19.6 MHz, 26.0 MHz, 32.4 MHz, and 38.8 MHz, respectively. In spite of the admonishments of Hranac Broadband (i.e., placement of 64-QAM channels only below 30 MHz, and using lower modulation order and lower bandwidth channels above 30 MHz to avoid undesirable diplex filter-related group delay effects), the channel plan 2000E shown in FIG. 21 represents the present maximum-capacity state of the art for actual implementations of upstream channel plans by some MSOs. From FIG. 21 and Table 2, and with the aid of Table 10 below, it can be seen that the channel plan 2000E has a deployed upstream capacity of just over approximately 120 Mbits/s (corresponding to an effective upstream capacity of approximately 100 Mbits/s using an 80% de-rating factor; see Chapman, page 8 and page 9, Table 1):
TABLE 10Channel BWRaw Data RateChannel QAM(MHz)(Mbits/s)646.430.72646.430.72646.430.72646.430.72TOTAL DEPLOYED CAPACITY122.88
Some industry commentators have indicated that an effective upstream capacity of approximately 100 Mbits/s provided by the channel plan 2000E of FIG. 21 is not so much a present reality as it is a near term target (e.g., see Chapman, pages 23 (Table 3), 63, and 138 regarding present-day maximum upstream capacity), at least in part due to ingress and other channel impairment issues discussed above (which arguably would impact the full functionality of at least the lowest frequency “leftmost” channel and the highest frequency “rightmost” channel of the set of four channels 2002P).
Recent trends toward improving upstream capacity in a cable communication system relate to using more effective DOCSIS Physical (PHY) layer technologies (e.g., S-CDMA, see Chapman, pages 88 through 95; Orthogonal Frequency Division Multiplexing or OFDM, see Chapman, pages 96 through 110), advanced error correction techniques (e.g., Low Density Parity Check or LDPC codes, see Chapman, pages 119 through 121), decreasing node and/or service group size (e.g., decreasing the number of subscriber premises per node and/or service group via node splitting or segmentation, see Chapman, pages 57 through 62), and expanding the range of frequencies allotted to the upstream path bandwidth (e.g., a “mid split” plan to expand the upstream path bandwidth to 85 MHz, a “high split” plan to expand the upstream path bandwidth to 200 MHz, and a “top-split” plan which would place additional upstream path bandwidth above 1 GHz) (e.g., see Chapman, pages 10 through 21; also see Al-Banna, throughout). One proposed channel plan relating to a “mid split” upstream path bandwidth expansion suggests the use of seven 256-QAM ATDMA 6.4 MHz bandwidth channels above 42 MHz (and no 256-QAM channels in the “original” upstream path bandwidth from 5 MHz to 42 MHz) (e.g., see Chapman, page 74, Table 17); this plan nonetheless requires S-CDMA channels below 20 MHz (e.g., as seen in the channel plans 2000C and 2000D of FIGS. 19 and 20, respectively).
Regarding proposals for expanding the range of frequencies allotted to the upstream path bandwidth, some commentators have noted various challenges with such an expansion; for example, from a system components perspective, diplexers would need to be changed throughout the hardline cable plant (such that the division between the upstream path bandwidth and the downstream path bandwidth would be moved to a higher portion of the spectrum), and active components as well as passive component may need to be retrofitted to accommodate higher frequency operation (see “HFC Network Capacity Expansion Options,” J. D. Salinger, The NCTA 2012 Spring Technical Forum Proceedings, May 21, 2012, hereafter “Salinger,” which publication is hereby incorporated herein by reference in its entirety). From an operational perspective, existing downstream analog channels would need to be removed, which may not be possible for many MSOs that are either required to maintain support for analog TVs directly, and/or are unable to remove analog channels for contractual reasons (Salinger, page 7). Even if removing analog channels is possible, this option appears to require the installation of filters in most or perhaps all subscriber premises equipment to both protect the new portion of the spectrum from emissions by existing subscriber equipment, and to protect existing equipment from transmissions by new subscriber premises equipment that would use the new portion of spectrum.